Merge pull request #642 from Routstr/fix/payout-safety-pool-lifecycle

Fix payout safety and proxy session lifecycle

.env.example

@@ -2,21 +2,51 @@
 UPSTREAM_BASE_URL=https://api.openai.com/v1
 UPSTREAM_API_KEY=your-upstream-api-key
 
-# ADMIN_PASSWORD=secure-admin-password
+# Tinfoil (confidential inference enclaves, EHBP)
+# TINFOIL_API_KEY=your-tinfoil-api-key
+
+# Secret key used to encrypt node secrets at rest (optional). If unset, the node
+# generates one on first start, writes it to routstr_secret.key (override the path
+# with ROUTSTR_SECRET_KEY_FILE), and prints it once — back that file up, because
+# losing the key makes previously encrypted secrets unreadable. Set it explicitly
+# to manage the key yourself (recommended in production). Generate one with:
+#   uv run python -c "from cryptography.fernet import Fernet; print(Fernet.generate_key().decode())"
+ROUTSTR_SECRET_KEY=
+
+# The admin password and the Nostr identity (nsec) are NOT set here. The admin
+# password is generated and logged once on first start (read it from the logs to
+# sign in); both are managed afterwards from the admin UI and stored encrypted in
+# the database. ADMIN_PASSWORD / NSEC are still read once as a legacy seed for
+# existing deployments, but new nodes should set them in the UI — a value left in
+# .env is ignored once the node has been configured.
 
 # Database
 # DATABASE_URL=sqlite+aiosqlite:///keys.db
+# Pool controls are validated at boot, sourced only from the environment, and
+# logged at startup. Keep total capacity across all workers below the database
+# connection limit. Pre-ping is automatic for networked backends; SQLite may
+# explicitly opt in if desired.
+# DATABASE_POOL_SIZE=5
+# DATABASE_MAX_OVERFLOW=10
+# DATABASE_POOL_TIMEOUT=30
+# DATABASE_POOL_RECYCLE=1800
+# DATABASE_POOL_PRE_PING=false
+# Warn when a checkout is held this many seconds.
+# DATABASE_POOL_HOLD_WARN_SECONDS=10
+# SQLite serialises writes; increasing its pool can trade pool timeouts for
+# "database is locked" errors rather than increasing write throughput.
 
 # Node Information
 # NAME=My Routstr Node
 # DESCRIPTION=Fast AI API access with Bitcoin payments
-# NSEC=nsec1...
 # HTTP_URL=https://api.mynode.com
 # ONION_URL=http://mynode.onion (auto fetched from compose)
 # RELAYS="wss://relay.damus.io,wss://relay.nostr.band,wss://eden.nostr.land,wss://relay.routstr.com"
 # ENABLE_ANALYTICS_SHARING=true
 # CASHU_MINTS="https://mint.minibits.cash/Bitcoin,https://mint.cubabitcoin.org,https://ecashmint.otrta.me"
+# MINT_OPERATION_CONCURRENCY=4
 # RECEIVE_LN_ADDRESS=
+# REFUND_SWEEP_CLAIM_TIMEOUT_SECONDS=900
 
 # Custom Pricing Configuration
 # MODEL_BASED_PRICING=true

.gitignore

@@ -1,6 +1,7 @@
 __pycache__
 .env
 keys.db
+routstr_secret.key
 wallet.sqlite3
 
 # Python build artifacts
@@ -16,6 +17,9 @@ dist/
 *.db-shm
 *.db-wal
 .*wallet.sqlite3
+.wallet/
+AGENTS.md
+TEST_SUITE_OVERVIEW.md
 *models.json
 .cashu
 .relay
@@ -38,3 +42,4 @@ proof_backups
 
 *.todo
 ui_out
+.worktrees

Dockerfile.full

@@ -7,7 +7,7 @@ WORKDIR /app/ui
 RUN corepack enable pnpm && corepack prepare pnpm@10.15.0 --activate
 
 # Copy UI source
-COPY ui/package.json ui/pnpm-lock.yaml* ./
+COPY ui/package.json ui/pnpm-lock.yaml* ui/pnpm-workspace.yaml* ./
 RUN pnpm install --frozen-lockfile
 
 COPY ui/ ./

Makefile

@@ -98,7 +98,7 @@ docker-down:
 lint:
 	@echo "🔍 Running linting checks..."
 	$(RUFF) check .
-	$(MYPY) routstr/ --ignore-missing-imports
+	$(MYPY) .
 
 format:
 	@echo "✨ Formatting code..."
@@ -107,7 +107,7 @@ format:
 
 type-check:
 	@echo "🔎 Running type checks..."
-	$(MYPY) routstr/ --ignore-missing-imports
+	$(MYPY) .
 
 # Development setup
 dev-setup:
@@ -234,7 +234,7 @@ ci-test:
 ci-lint:
 	@echo "🤖 Running CI linting..."
 	$(RUFF) check . --exit-non-zero-on-fix
-	$(MYPY) routstr/ --ignore-missing-imports --no-error-summary
+	$(MYPY) . --no-error-summary
 
 # Debug helpers
 test-debug:

README.md

@@ -55,19 +55,41 @@ If you are a node runner, start a Routstr Core instance using Docker Compose:
 
 1. **Prepare your `.env`**:
    ```bash
-   ADMIN_PASSWORD=mysecretpassword
+   # Optional: encrypts node secrets at rest. If unset, the node generates a key
+   # on first start, writes it to routstr_secret.key, and prints it once — back
+   # up that file. Set it explicitly to manage the key yourself (recommended in
+   # production).
+   ROUTSTR_SECRET_KEY=<generated-key>
    NAME="My AI Node"
    DESCRIPTION="Fast access to models"
-   NSEC=yournsec
    RECEIVE_LN_ADDRESS=yourname@wallet.com
    ```
 
+   Your Nostr identity (`nsec`) is not set in `.env` — configure it from the admin
+   UI after first start, where it's stored encrypted in the database. (`NSEC` in
+   `.env` is still read once as a legacy seed for existing deployments.)
+
+   If you don't set one, a key is generated and printed on first start — save it
+   somewhere safe (losing it makes previously encrypted secrets unreadable). To
+   supply your own, generate it once and keep it stable:
+   ```bash
+   uv run python -c "from cryptography.fernet import Fernet; print(Fernet.generate_key().decode())"
+   ```
+
 2. **Start the services**:
    ```bash
    docker compose up -d
    ```
 
-3. **Configure**:
+3. **Get your admin password**:
+   On first start the node generates an admin password and logs it once with the
+   `/admin` URL. Read it from the logs:
+   ```bash
+   docker compose logs routstr | grep -i admin
+   ```
+   (Lost it? Reset with `docker compose exec routstr /.venv/bin/python scripts/reset_admin_password.py --regenerate`.)
+
+4. **Configure**:
    Open [http://localhost:8000/admin/](http://localhost:8000/admin/) to connect your AI providers and set pricing.
 
 For full instructions, see the **[Provider Quick Start Guide](https://docs.routstr.com/provider/quickstart/)**.

compose.yml

@@ -13,6 +13,7 @@ services:
       - ./ui_out:/output:z
     command:
       ["sh", "-c", "mkdir -p /output && cp -r /app/built/. /output/ && echo 'UI build copied to mounted volume' && ls -la /output/ && echo 'UI built and ready' && tail -f /dev/null"]
+    restart: unless-stopped
 
   routstr:
     build: .
@@ -31,6 +32,7 @@ services:
       - 8000:8000
     extra_hosts: # Needed to access locally running models
       - "host.docker.internal:host-gateway"
+    restart: unless-stopped
 
   tor:
     image: ghcr.io/hundehausen/tor-hidden-service:latest
@@ -41,6 +43,7 @@ services:
       - HS_ROUTER=routstr:8000:80
     depends_on:
       - routstr
+    restart: unless-stopped
 
 volumes:
   tor-data:

docs/api/errors.md

@@ -113,42 +113,109 @@ All errors follow a consistent JSON structure:
 **Status:** 402  
 **Resolution:** Top up API key balance
 
-#### Invalid Token
+### Cashu Token Redemption Errors
+
+These errors are returned when a Cashu token you pay with cannot be redeemed.
+They apply to every endpoint that accepts a token:
+
+- **Per-request payment** via the `X-Cashu` header (chat completions + Responses API).
+- **API key top-up** via `POST /v1/wallet/topup`.
+- **Minting an API key** from a token sent in `Authorization: Bearer <cashu-token>`.
+
+All three share one classifier, so the same failure yields the same HTTP status
+and sanitized message everywhere. Structured error envelopes (`X-Cashu` and
+`Authorization: Bearer <cashu-token>`) also expose the same `type` and `code` —
+branch on `type` (or `code` for finer granularity). `POST /v1/wallet/topup`
+keeps its existing plain-string `detail` envelope, so branch on status there.
+
+| `type` | Status | `code` | Retryable | Meaning |
+|--------|--------|--------|-----------|---------|
+| `token_already_spent` | 400 | `cashu_token_already_spent` | No | The token was already redeemed. |
+| `invalid_token` | 400 | `invalid_cashu_token` | No | The token is malformed or cannot be decoded. |
+| `mint_error` | 422 | `cashu_token_swap_fees_exceed_amount` | No | Token value is too small to cover the mint's swap/melt fees. |
+| `mint_error` | 422 | `cashu_foreign_mint_swap_failed` | No | Swapping the token from a foreign mint to the primary mint failed. |
+| `mint_unreachable` | 503 | `cashu_mint_unreachable` | **Yes** | The mint could not be reached (DNS failure, refused/reset connection, timeout). The token is fine — retry once the mint recovers. |
+| `cashu_error` | 400 | `cashu_token_redemption_failed` | No | The token could not be redeemed for another expected reason. |
+| `cashu_error` | 400 | `cashu_token_zero_value` | No | The token redeemed to zero (empty/dust token, or value fully consumed by fees). |
+| `token_consumed` | 500 | `cashu_token_consumed` | No | The token was **spent** (melted/redeemed) but crediting it then failed. Do not retry — the token is gone; contact support to reconcile. |
+| `api_error` | 500 | `internal_error` | Maybe | Unexpected server-side fault during redemption. |
+
+!!! important "Retry only `mint_unreachable`"
+    Only `mint_unreachable` (503) means the same token will work again later —
+    everything else is a permanent property of the token and must not be
+    blindly retried. Use exponential backoff for the 503. In particular, a
+    `token_consumed` 500 means the mint already spent the token, so a retry
+    would fail as `token_already_spent`.
+
+#### Mint Unreachable (retryable)
 
 ```json
 {
   "error": {
-    "type": "payment_error",
-    "message": "Invalid Cashu token",
-    "code": "invalid_token",
-    "details": {
-      "reason": "Token already spent"
-    }
+    "type": "mint_unreachable",
+    "message": "Cashu mint is unreachable",
+    "code": "cashu_mint_unreachable"
   }
 }
 ```
 
-**Status:** 400  
-**Resolution:** Use a valid, unspent token
+**Status:** 503
 
-#### Mint Unavailable
+**Resolution:** The token is valid — the mint is temporarily down. Retry with
+backoff, or pay with a token from a different mint.
+
+#### Token Already Spent
 
 ```json
 {
   "error": {
-    "type": "payment_error",
-    "message": "Cannot connect to Cashu mint",
-    "code": "mint_unavailable",
-    "details": {
-      "mint_url": "https://mint.example.com",
-      "retry_after": 60
-    }
+    "type": "token_already_spent",
+    "message": "Cashu token already spent",
+    "code": "cashu_token_already_spent"
   }
 }
 ```
 
-**Status:** 503  
-**Resolution:** Try again later or use different mint
+**Status:** 400
+
+**Resolution:** Use a fresh, unspent token. Do not retry with the same token.
+
+#### Response envelope differs by endpoint
+
+The `error` object above is identical everywhere, but the surrounding envelope
+depends on how you paid:
+
+- **`X-Cashu` header payments** (chat + Responses API) return the object at the
+  top level, alongside a `request_id`:
+
+  ```json
+  {
+    "error": { "type": "mint_unreachable", "message": "Cashu mint is unreachable", "code": "cashu_mint_unreachable" },
+    "request_id": "req-abc123"
+  }
+  ```
+
+  The original token is echoed back in the `X-Cashu` **response header only when
+  it is still spendable** (e.g. `mint_unreachable`, `invalid_cashu_token`, fee
+  errors) so you can recover/retry it. It is **not** echoed for spent/consumed
+  tokens (`cashu_token_already_spent`, `cashu_token_consumed`,
+  `cashu_token_zero_value`, `internal_error`) — retrying those can never succeed.
+
+- **`Authorization: Bearer <cashu-token>`** (API key minting) wraps it in
+  FastAPI's `detail` field:
+
+  ```json
+  { "detail": { "error": { "type": "mint_unreachable", "message": "Cashu mint is unreachable", "code": "cashu_mint_unreachable" } } }
+  ```
+
+- **`POST /v1/wallet/topup`** returns a plain string message under `detail` —
+  it carries the shared HTTP **status** and **message** (e.g. `503` for an
+  unreachable mint) but not the structured `type`/`code`, so branch on the
+  status code here:
+
+  ```json
+  { "detail": "Cashu mint is unreachable" }
+  ```
 
 ### Validation Errors
 
@@ -347,7 +414,7 @@ class ErrorHandler:
         'rate_limit',
         'upstream_timeout',
         'model_overloaded',
-        'mint_unavailable'
+        'cashu_mint_unreachable'
     }
     
     # Errors requiring user action

docs/ehbp-proxy-support.md

@@ -0,0 +1,150 @@
+# EHBP Proxy Support for Tinfoil Models
+
+## Problem
+
+The SDK's `SecureClient.fetch` encrypts request bodies with HPKE (EHBP protocol)
+and sends them to the Routstr provider. The Routstr proxy had no EHBP handling:
+
+1. It tried to `json.loads()` the binary HPKE-sealed body → failed with a 400
+2. The upstream PPQ.AI public endpoint (`/v1/chat/completions`) doesn't speak
+   EHBP, so the response had no `Ehbp-Response-Nonce` header
+3. `SecureClient` threw `Missing Ehbp-Response-Nonce header` because it expects
+   every response from an EHBP-configured `baseURL` to carry that header
+
+## Root cause
+
+PPQ.AI exposes EHBP-aware inference at `/private/`, separate from the public
+`/v1/` endpoint. The Routstr proxy was forwarding to `/v1/` (the public
+endpoint) instead of `/private/` (the enclave endpoint). The public endpoint
+can't decrypt the body, returns a normal HTTP response, and the SDK can't
+decrypt it because there's no nonce header.
+
+The PPQ private-mode proxy (`ppq-private-mode-proxy/lib/proxy.ts`) shows the
+correct pattern: `SecureClient` talks to `api.ppq.ai/private/v1/chat/completions`,
+which decrypts inside the attested enclave and returns an EHBP-encrypted
+response with the `Ehbp-Response-Nonce` header.
+
+## What was changed
+
+### `routstr/proxy.py`
+
+Detects EHBP requests by checking for the `Ehbp-Encapsulated-Key` header (set
+by the EHBP transport on every encrypted request). For EHBP requests:
+
+- Skips JSON body parsing (the body is binary ciphertext, not JSON)
+- Reads the model ID from the `X-Routstr-Model` header (set by the SDK) instead
+  of from `body.model`
+- Routes through new `forward_ehbp_request` (bearer auth) and
+  `forward_ehbp_x_cashu_request` (x-cashu auth) methods
+- Skips reactive 400 param correction (can't parse encrypted response body)
+- Still charges the user via `pay_for_request` (uses `max_cost_for_model` from
+  the model registry, not the body)
+
+### `routstr/upstream/base.py`
+
+Keeps EHBP as an explicit opt-in provider capability instead of making every
+upstream provider appear EHBP-capable:
+
+- `supports_ehbp = False` by default
+- `get_ehbp_forwarding_target(path, model_obj)` raises `NotImplementedError`
+  unless a provider opts in and returns a provider-specific EHBP target
+
+The actual EHBP forwarding logic does **not** live in `base.py`.
+
+### `routstr/upstream/ehbp.py`
+
+Contains the shared opaque EHBP transport and billing helpers:
+
+- `EHBPForwardingTarget` — provider-specific target URL plus extra headers
+- `forward_ehbp_request()` — forwards the encrypted body, captures Tinfoil
+  usage from a response header or streaming HTTP trailer, and finalizes bearer
+  billing at actual cost (falling back to max cost when usage is unavailable)
+- `forward_ehbp_x_cashu_request()` — redeems the Cashu token, refunds the full
+  token on upstream failure, and refunds the difference between the redeemed
+  amount and actual cost (or max cost when usage is unavailable)
+
+### Provider support
+
+EHBP is currently enabled only for `TinfoilUpstreamProvider`. It forwards to
+Tinfoil's attested enclave and requests `X-Tinfoil-Usage-Metrics` for billing.
+PPQ.AI retains its private-target implementation, but `supports_ehbp = False`
+until it has a provider-specific trusted usage/model-binding strategy.
+
+## Why it's done this way
+
+The proxy is a **blind relay** for EHBP requests. It cannot decrypt the body
+(only the attested enclave can), so it must:
+
+1. Get the model ID from a header, not the body
+2. Forward the raw bytes without parsing or transformation
+3. Stream the response back without SSE/cost parsing
+4. Pass through EHBP protocol headers (`Ehbp-Encapsulated-Key` on request,
+   `Ehbp-Response-Nonce` on response)
+
+Cost tracking happens at the proxy level. Routstr reserves or redeems up to
+`max_cost_for_model`, then Tinfoil's out-of-band usage header/trailer allows it
+to finalize at actual token cost. If trusted usage is missing or invalid, the
+proxy safely falls back to max-cost billing.
+
+## End-to-end flow
+
+```
+SDK                                  Routstr Proxy                    PPQ.AI /private/
+  │                                       │                                │
+  │── X-Routstr-Model: tinfoil-kimi-k2-6 ─│                                │
+  │── Ehbp-Encapsulated-Key: <hex> ───────│                                │
+  │── Authorization: Bearer <cashu> ──────│                                │
+  │── body = HPKE-encrypted(kimi-k2-6) ───│                                │
+  │                                       │                                │
+  │                                  detects Ehbp-Encapsulated-Key         │
+  │                                  reads model from X-Routstr-Model      │
+  │                                  does billing/routing                  │
+  │                                       │                                │
+  │                                  adds X-Private-Model: private/kimi-k2-6
+  │                                  forwards raw body to /private/v1/...  │
+  │                                       │──────────────────────────────▶│
+  │                                       │                          enclave decrypts
+  │                                       │                          runs inference
+  │                                       │◀── Ehbp-Response-Nonce ──────│
+  │                                       │◀── encrypted response ────────│
+  │                                       │                                │
+  │                                  streams response back untouched       │
+  │◀── encrypted response ────────────────│                                │
+  │                                        │                                │
+  SecureClient reads nonce, decrypts       │                                │
+  SDK SSE processing sees plaintext        │                                │
+```
+
+## Model ID mapping
+
+Three parties see three different model IDs:
+
+| Party | Header/Body | Value | Source |
+|---|---|---|---|
+| Routstr proxy | `X-Routstr-Model` header | `tinfoil-kimi-k2-6` | SDK sends full caller-facing id |
+| Tinfoil usage metrics | `model` field | `kimi-k2-6` | Enclave reports the model actually served |
+| Tinfoil enclave | `body.model` (encrypted) | `kimi-k2-6` | SDK strips `tinfoil-` prefix before encryption |
+
+## Implementation status
+
+A dedicated `TinfoilUpstreamProvider` (`routstr/upstream/tinfoil.py`) now
+implements the direct blind-upstream pattern described above. The shared EHBP
+helpers in `routstr/upstream/ehbp.py` were extended to:
+
+- Request usage metrics via `X-Tinfoil-Request-Usage-Metrics: true`.
+- Parse `X-Tinfoil-Usage-Metrics` from the response header (non-streaming) or
+  HTTP trailer (streaming).
+- Override the forwarding URL with a validated `X-Tinfoil-Enclave-Url` when the
+  SDK sends it.
+- Finalize bearer billing with the dedicated EHBP actual-cost finalizer.
+- Compute X-Cashu refunds from actual cost instead of max cost.
+
+See `docs/tinfoil-direct-integration.md` for the full implementation notes.
+
+## Verification status
+
+Unit coverage includes usage parsing, target validation, HTTP trailer capture,
+response-size limits, and bearer payment finalization. End-to-end requests have
+verified both non-streaming usage headers and streaming usage trailers against
+Tinfoil. SDK behavior on proxy-generated non-2xx responses without an
+`Ehbp-Response-Nonce` still merits explicit end-to-end coverage.

docs/provider/configuration.md

@@ -13,7 +13,10 @@ Before running your node, you should create a `.env` file in the project root. T
 ### Example .env
 
 ```bash
-ADMIN_PASSWORD=your-secure-password
+# Encrypts node secrets at rest. Optional — if unset, the node generates a key on
+# first start and prints it once (back it up). Set it to manage the key yourself
+# (recommended in production). See "Secrets at Rest" below.
+ROUTSTR_SECRET_KEY=
 
 # Node Identity
 NAME="My AI Node"
@@ -25,10 +28,10 @@ RECEIVE_LN_ADDRESS=yourname@wallet.com
 
 ### Setting the UI Password
 
-There are two ways to set or change your Admin Dashboard password:
+On first start the node generates an admin password and logs it once — read it from the container logs to sign in. You can then change it two ways:
 
-1.  **Via Environment Variable**: Set `ADMIN_PASSWORD` in your `.env` file before starting the container. This will be the password used for the first login.
-2.  **Via Dashboard**: Once logged in, go to **Settings** → **Security** to update your password. Dashboard settings override the `.env` file once saved.
+1.  **Via Dashboard**: Once logged in, go to **Settings** → **Security** to update your password.
+2.  **Via Environment Variable (legacy seed)**: Setting `ADMIN_PASSWORD` in `.env` before the first start seeds the initial password instead of generating one. It's read only once, for existing deployments; a value left in `.env` is ignored after the node has been configured.
 
 ---
 
@@ -123,12 +126,14 @@ Use environment variables for:
 | -------------------- | --------------------------------- | ------------------------------------ |
 | `UPSTREAM_BASE_URL`  | Upstream API endpoint             | —                                    |
 | `UPSTREAM_API_KEY`   | Upstream API key                  | —                                    |
-| `ADMIN_PASSWORD`     | Dashboard password                | (none)                               |
+| `ADMIN_PASSWORD`     | Legacy seed for the dashboard password (otherwise generated + logged on first start) | (auto-generated) |
+| `ROUTSTR_SECRET_KEY` | Master key encrypting node secrets at rest. Auto-generated to a key file if unset | (auto-generated) |
+| `ROUTSTR_SECRET_KEY_FILE` | Path to the generated key file (used when `ROUTSTR_SECRET_KEY` is unset) | `routstr_secret.key` beside the database |
 | `DATABASE_URL`       | Database connection string        | `sqlite+aiosqlite:///keys.db`        |
 | `NAME`               | Node display name                 | `ARoutstrNode`                       |
 | `DESCRIPTION`        | Node description                  | `A Routstr Node`                     |
 | `NPUB`               | Nostr public key (bech32)         | —                                    |
-| `NSEC`               | Nostr private key                 | —                                    |
+| `NSEC`               | Legacy seed for the Nostr private key (otherwise set from the admin UI) | —                |
 | `ENABLE_ANALYTICS_SHARING` | Enable usage analytics sharing to Nostr | `true`                         |
 | `CASHU_MINTS`        | Comma-separated mint URLs         | `https://mint.minibits.cash/Bitcoin` |
 | `RECEIVE_LN_ADDRESS` | Lightning address for withdrawals | —                                    |
@@ -142,6 +147,20 @@ Use environment variables for:
 
 Environment variables are read on startup. Dashboard settings override them and persist in the database. Once you change a setting in the dashboard, the env var is ignored for that setting.
 
+### Secrets at Rest
+
+The node's Nostr private key (`nsec`) is encrypted in the database using
+`ROUTSTR_SECRET_KEY`. You don't have to set it: if it's unset, the node generates a
+key on first start, writes it **beside the database** (the file named by
+`ROUTSTR_SECRET_KEY_FILE`, default `routstr_secret.key`) so it persists on the same
+volume as your data, and prints it once.
+
+**Back up that key** — it lives on the same volume as your database, so include it
+in your backups. If it is lost or changed, previously encrypted secrets can't be
+decrypted and must be re-entered — there is no rotation. To keep the key off the
+data volume, set `ROUTSTR_SECRET_KEY` explicitly (an env value always takes
+precedence over the file). See also [Deployment](deployment.md).
+
 ---
 
 ## Models

docs/provider/deployment.md

@@ -38,7 +38,6 @@ services:
       - routstr-data:/app/data
     environment:
       DATABASE_URL: "sqlite:////app/data/routstr.db"
-      ADMIN_KEY: "your-secure-admin-key"
       LOG_LEVEL: "info"
 
 volumes:
@@ -87,6 +86,8 @@ services:
       - ./logs:/app/logs
     environment:
       - TOR_PROXY_URL=socks5://tor:9050
+      # Keep the database (and the key file generated beside it) on the volume.
+      - DATABASE_URL=sqlite:////app/data/routstr.db
     depends_on:
       - tor
 
@@ -125,8 +126,8 @@ services:
       - UPSTREAM_BASE_URL=https://api.openai.com/v1
       - UPSTREAM_API_KEY=sk-proj-...
       
-      # Secure the dashboard (recommended)
-      - ADMIN_PASSWORD=your-secure-password
+      # The admin password is generated and logged once on first start; set
+      # ADMIN_PASSWORD here only as a legacy seed for an existing deployment.
       
       # Node identity
       - NAME=My Provider Node
@@ -134,6 +135,9 @@ services:
       
       # Lightning withdrawals
       - RECEIVE_LN_ADDRESS=me@walletofsatoshi.com
+
+      # Keep the database (and the key file generated beside it) on the volume.
+      - DATABASE_URL=sqlite:////app/data/routstr.db
     volumes:
       - ./data:/app/data
 ```
@@ -155,22 +159,36 @@ Example `.env`:
 ```bash
 UPSTREAM_BASE_URL=https://api.openai.com/v1
 UPSTREAM_API_KEY=sk-proj-...
-ADMIN_PASSWORD=change-me
+# Keep the database (and the key file generated beside it) on the mounted volume.
+DATABASE_URL=sqlite:////app/data/routstr.db
+# Encrypts node secrets at rest. Optional — if unset, a key is generated next to
+# your database (on the same volume) and its file is named once for backup. Set
+# it explicitly to manage the key yourself.
+ROUTSTR_SECRET_KEY=
 NAME=My Provider Node
 RECEIVE_LN_ADDRESS=me@walletofsatoshi.com
 ```
 
+!!! note "Secret key persistence"
+    If you leave `ROUTSTR_SECRET_KEY` unset, the node generates one and stores it
+    as `routstr_secret.key` **next to your database**, so it persists on the same
+    volume as your data — just include that volume in your backups. For stronger
+    isolation (keeping the key off the data volume), set `ROUTSTR_SECRET_KEY` from
+    a secrets manager instead.
+
 See [Configuration](configuration.md) for all available options.
 
 ---
 
 ## Persistence
 
-Routstr stores all data in `/app/data`:
+Point `DATABASE_URL` inside `/app/data` (as the examples above do) so everything
+Routstr persists lands on the mounted volume:
 
 | Path | Contents |
 |------|----------|
-| `keys.db` | SQLite database (settings, API keys, sessions) |
+| `routstr.db` | SQLite database (settings, API keys, sessions) |
+| `routstr_secret.key` | Auto-generated master key, written beside the database when `ROUTSTR_SECRET_KEY` is unset |
 | `.wallet/` | Cashu wallet data (your Bitcoin!) |
 
 !!! warning "Back Up Your Data"

docs/provider/quickstart.md

@@ -29,19 +29,25 @@ In future versions, you'll be able to run a node that connects to other Routstr
 Create a `.env` file in the root of the project to store your secrets:
 
 ```bash
-# Initial Admin Password
-ADMIN_PASSWORD=mysecretpassword
+# Encrypts node secrets at rest. Optional — if unset, the node generates a key on
+# first start and prints it once (back it up).
+ROUTSTR_SECRET_KEY=
 
 # Node Identity
 NAME="My AI Node"
 DESCRIPTION="Fast access to models"
-NSEC=yournsec
 
 # Lightning Payouts
 RECEIVE_LN_ADDRESS=yourname@wallet.com
 
 ```
 
+The admin password is generated and logged once on first start (read it from the
+logs to sign in), and your Nostr identity (`nsec`) is configured afterwards from
+the admin UI — both are stored encrypted in the database, not in `.env`.
+(`ADMIN_PASSWORD` / `NSEC` are still read once as a legacy seed for existing
+deployments.)
+
 ## 2. Start the Node
 
 The recommended way to run Routstr is using Docker Compose, which handles the node, the UI, and optional services like Tor.
@@ -72,7 +78,7 @@ docker compose up -d
 Open the **Admin Dashboard** at [http://localhost:8000/admin/](http://localhost:8000/admin/).
 
 !!! note "Login"
-Use the `ADMIN_PASSWORD` you defined in your `.env` file to log in. If you didn't set one, the dashboard will prompt you to set one on first visit.
+On first start the node generates an admin password and logs it once — read it from the container logs to sign in. You can change it afterwards from **Settings** → **Security**.
 
 ### Connect Your AI Providers
 

docs/tinfoil-direct-integration.md

@@ -0,0 +1,493 @@
+# Tinfoil / PPQ Private-Mode Integration Notes
+
+This document summarizes the current options for integrating Tinfoil/PPQ private models with Routstr, based on the EHBP work in this branch and local testing against `ppq-private-mode-proxy`.
+
+## Background
+
+PPQ private models run behind a Tinfoil/EHBP flow:
+
+- Request bodies are HPKE-encrypted by a Tinfoil client.
+- The PPQ `/private/` endpoint routes ciphertext to the attested enclave.
+- The enclave decrypts, runs inference, and returns an encrypted response.
+- The caller's Tinfoil client decrypts the response locally.
+
+The PPQ private-mode proxy (`~/projects/ppq-private-mode-proxy`) uses this pattern with the JavaScript `tinfoil` SDK:
+
+```ts
+import { SecureClient } from "tinfoil";
+
+const apiBase = "https://api.ppq.ai";
+
+const client = new SecureClient({
+  baseURL: `${apiBase}/private/`,
+  attestationBundleURL: `${apiBase}/private`,
+  transport: "ehbp",
+});
+
+await client.ready();
+
+const response = await client.fetch(`${apiBase}/private/v1/chat/completions`, {
+  method: "POST",
+  headers: {
+    "Content-Type": "application/json",
+    Authorization: `Bearer ${process.env.PPQ_API_KEY}`,
+    "X-Private-Model": "private/gpt-oss-120b",
+    "x-query-source": "api",
+  },
+  body: JSON.stringify({
+    model: "gpt-oss-120b", // enclave-internal model id
+    messages: [{ role: "user", content: "Hello" }],
+  }),
+});
+
+const json = await response.json();
+console.log(json.usage);
+```
+
+`X-Private-Model` carries the PPQ-facing private model id, while the encrypted JSON body uses the enclave-internal model id without the `private/` prefix.
+
+## PPQ private model pricing
+
+PPQ exposes private model pricing through:
+
+```text
+GET https://api.ppq.ai/v1/models?type=all
+```
+
+Filter models whose IDs start with `private/`.
+
+Example private pricing observed:
+
+| Model | Input USD / 1M tokens | Output USD / 1M tokens |
+|---|---:|---:|
+| `private/gpt-oss-120b` | `0.79125` | `1.31875` |
+| `private/llama3-3-70b` | `1.84625` | `2.90125` |
+| `private/qwen3-vl-30b` | `1.31875` | `4.22` |
+| `private/glm-5-2` | `1.5825` | `5.53875` |
+| `private/gemma4-31b` | `0.47475` | `1.055` |
+| `private/kimi-k2-6` | `1.5825` | `5.53875` |
+
+The `ppq-private-mode-proxy` commit `ba984214793d3bca0f7d046b6955d42abc1c6843` changed OpenClaw display metadata to align with a 5% API margin. The live PPQ model endpoint returns the more precise rates above, which already include that margin.
+
+Actual PPQ private billing is:
+
+```text
+price_usd =
+  input_tokens  * input_per_1M_tokens  / 1_000_000
++ output_tokens * output_per_1M_tokens / 1_000_000
+```
+
+A test request to `private/gpt-oss-120b` produced:
+
+```json
+{
+  "input_count": 74,
+  "output_count": 3,
+  "price_in_usd": 0.00006250875
+}
+```
+
+which exactly matches:
+
+```text
+74 * 0.79125 / 1_000_000 + 3 * 1.31875 / 1_000_000
+= 0.00006250875 USD
+```
+
+## Integration architectures
+
+There are three materially different ways Routstr could integrate Tinfoil/PPQ private inference.
+
+## Option A: User integrates Tinfoil directly
+
+```text
+User app / SDK
+  -> Tinfoil SecureClient / TinfoilAI
+    -> EHBP-encrypted request
+      -> Tinfoil/PPQ private enclave
+```
+
+Properties:
+
+- Best privacy for the user.
+- Routstr is not in the request path.
+- User's client encrypts requests and decrypts responses.
+- Usage is visible to the user's app after decryption.
+- Billing is handled directly by Tinfoil/PPQ.
+
+Example with Tinfoil's OpenAI-compatible client:
+
+```ts
+import { TinfoilAI } from "tinfoil";
+
+const client = new TinfoilAI({
+  apiKey: process.env.TINFOIL_API_KEY,
+  transport: "ehbp",
+});
+
+const res = await client.chat.completions.create({
+  model: "llama3-3-70b",
+  messages: [{ role: "user", content: "Hello" }],
+});
+
+console.log(res.choices[0].message.content);
+console.log(res.usage);
+```
+
+This is not a Routstr marketplace flow unless Routstr only acts as discovery/UI around direct Tinfoil/PPQ usage.
+
+## Option B: Routstr integrates Tinfoil as an upstream client
+
+```text
+User -> Routstr plaintext request
+  -> Routstr Tinfoil SecureClient encrypts to PPQ/Tinfoil
+    -> PPQ private enclave
+  -> Routstr receives decrypted response
+  -> Routstr bills from decrypted usage
+  -> Routstr returns plaintext response to user
+```
+
+Properties:
+
+- Easier exact billing.
+- Routstr can read the decrypted OpenAI response and `usage` object.
+- Routstr can charge exact PPQ token pricing.
+- Privacy is different: the user sends plaintext to Routstr, and Routstr sees prompts/responses.
+- End-to-end encryption is only Routstr-to-enclave, not user-to-enclave.
+
+This should be considered a separate product/provider mode, not the same as an end-to-end private relay.
+
+Practical implementation options:
+
+1. Run a Node sidecar that uses the `tinfoil` npm package and expose it as a local HTTP upstream to Routstr.
+2. Port EHBP client behavior to Python.
+3. Reuse or adapt `ppq-private-mode-proxy` as a local upstream.
+
+A Node sidecar is probably the quickest implementation path because `ppq-private-mode-proxy` already demonstrates the full flow.
+
+## Option C: Routstr uses Tinfoil as a direct blind upstream
+
+This is the current branch's design intent and is likely the best fit if Tinfoil/PPQ exposes usage metadata headers:
+
+```text
+User Tinfoil SecureClient
+  -> encrypted request body
+    -> Routstr proxy
+      -> Tinfoil/PPQ private API
+         with X-Tinfoil-Request-Usage-Metrics: true
+        -> PPQ private enclave
+      <- encrypted response
+         plus X-Tinfoil-Usage-Metrics / cost headers
+    <- Routstr proxy
+  -> user decrypts response
+```
+
+In this mode Routstr is still a normal upstream proxy from the user's point of view, but the upstream is Tinfoil/PPQ private inference and the body remains opaque to Routstr.
+
+Properties:
+
+- Strongest privacy with Routstr in the path.
+- Routstr never sees plaintext prompt or plaintext response.
+- Routstr can authenticate and route based on plaintext headers.
+- Routstr should request Tinfoil usage metadata by adding `X-Tinfoil-Request-Usage-Metrics: true` to the upstream request.
+- Tinfoil documents `X-Tinfoil-Usage-Metrics` as an upstream response header for non-streaming requests.
+- For streaming requests, Tinfoil documents `X-Tinfoil-Usage-Metrics` as an HTTP trailer available only after the response body completes.
+- If Tinfoil/PPQ returns `X-Tinfoil-Usage-Metrics` or a cost header on the encrypted response, Routstr can bill exactly without decrypting the body.
+- If usage is only present inside the encrypted response body, Routstr still cannot read it and exact billing is not possible without a separate metadata path.
+
+This is the only architecture that preserves end-to-end encryption from the user to the PPQ/Tinfoil enclave while still letting Routstr mediate payment. The key requirement is that usage/cost metadata must be returned outside the encrypted body, ideally as a response header available before body streaming begins.
+
+## Current Routstr problem
+
+The current EHBP implementation charges successful EHBP requests at `max_cost_for_model` because Routstr cannot decrypt the response body:
+
+```text
+successful EHBP request -> charge full reserved max cost
+```
+
+That is incorrect for PPQ private models. Max cost should be only a reservation/solvency ceiling. Final charge should use actual PPQ private token pricing.
+
+Desired behavior:
+
+```text
+reserve max cost
+forward encrypted request
+obtain actual usage/cost metadata
+finalize actual cost
+refund/release the difference
+```
+
+## Usage/cost metadata requirement
+
+For blind-relay exact billing, PPQ should return one of the following outside the encrypted response body:
+
+```http
+X-PPQ-Cost-USD: 0.00006250875
+```
+
+or:
+
+```http
+X-Private-Usage-Metrics: input=74,output=3
+```
+
+or:
+
+```http
+X-Tinfoil-Usage-Metrics: prompt=74,completion=3,total=77
+```
+
+Tinfoil proxy documentation references a usage-metrics flow where a proxy can request usage via:
+
+```http
+X-Tinfoil-Request-Usage-Metrics: true
+```
+
+and read usage from:
+
+```http
+X-Tinfoil-Usage-Metrics
+```
+
+Docs/example references:
+
+- https://docs.tinfoil.sh/guides/proxy-server
+- https://github.com/tinfoilsh/encrypted-request-proxy-example
+
+During local PPQ testing, PPQ responses included this CORS exposure header:
+
+```http
+Access-Control-Expose-Headers: Ehbp-Response-Nonce, X-Private-Usage-Metrics, X-Encrypted-Usage-Metrics, X-Tinfoil-Usage-Metrics
+```
+
+However, the actual tested non-streaming response did not include any of these usage headers, even when `X-Tinfoil-Request-Usage-Metrics: true` was sent.
+
+The decrypted body did include normal OpenAI usage, but only the decrypting Tinfoil client can see that body.
+
+## Query-history fallback
+
+PPQ's query history endpoint exposes actual usage and cost:
+
+```text
+GET https://api.ppq.ai/queries/history?page=1&page_count=...
+```
+
+A record includes:
+
+```json
+{
+  "timestamp": "...",
+  "model": "private/gpt-oss-120b",
+  "input_count": 74,
+  "output_count": 3,
+  "price_in_usd": 0.00006250875,
+  "query_type": "chat_completion",
+  "query_source": "api"
+}
+```
+
+This could be used as a fallback, but it is less robust than response headers/trailers because matching a request to a history row can be race-prone under concurrency. It would need a reliable request identifier or metadata field that PPQ stores in history.
+
+## Recommended Routstr direction
+
+Use Tinfoil/PPQ as a direct blind upstream and have Routstr explicitly request usage metadata:
+
+```text
+User encrypts body
+Routstr reserves max cost
+Routstr forwards encrypted body to Tinfoil/PPQ /private/
+Routstr includes X-Tinfoil-Request-Usage-Metrics: true
+Tinfoil/PPQ returns X-Tinfoil-Usage-Metrics as a response header for non-streaming,
+or as an HTTP trailer after the body completes for streaming
+Routstr finalizes exact charge
+User decrypts encrypted response
+```
+
+This preserves both:
+
+- privacy: Routstr cannot read prompts/responses;
+- exact billing: Routstr can charge actual PPQ private model cost, assuming Tinfoil/PPQ returns usage/cost metadata outside the encrypted body.
+
+Implementation steps:
+
+1. Update PPQ model fetching to include private models:
+
+   ```text
+   GET https://api.ppq.ai/v1/models?type=all
+   ```
+
+2. Register `private/*` models and any Routstr-facing aliases with correct `forwarded_model_id`.
+
+3. Keep max-cost reservation for bearer keys and X-Cashu solvency checks.
+
+4. Add parsing support for possible usage/cost headers:
+
+   ```http
+   X-PPQ-Cost-USD
+   X-Private-Usage-Metrics
+   X-Tinfoil-Usage-Metrics
+   X-Encrypted-Usage-Metrics
+   ```
+
+5. Finalize by actual cost instead of max cost:
+
+   ```text
+   actual_msats = ceil((actual_usd / sats_usd_price()) * 1000)
+   actual_msats = max(actual_msats, settings.min_request_msat)
+   actual_msats = min(actual_msats, reserved_msats)
+   ```
+
+   or, if only token counts are available:
+
+   ```text
+   actual_usd =
+     input_tokens  * input_per_1M_tokens  / 1_000_000
+   + output_tokens * output_per_1M_tokens / 1_000_000
+   ```
+
+6. If usage/cost metadata is missing, choose an explicit policy:
+
+   - fail closed and refund/revert;
+   - query PPQ history as a fallback;
+   - fallback to max-cost billing only if explicitly configured and clearly disclosed.
+
+Silent max-cost billing should not be the default for PPQ private requests.
+
+## X-Cashu consideration
+
+For bearer-auth requests, finalization can happen after the response stream completes if usage is delivered as a trailer.
+
+For `X-Cashu`, Routstr needs to return the refund token in the response headers. If usage/cost is only available after consuming the encrypted response stream, Routstr may need to buffer EHBP responses before sending them to the client so it can compute the refund amount first.
+
+Possible approaches:
+
+1. Prefer a non-trailer response header with actual cost, available before streaming body starts.
+2. Buffer EHBP X-Cashu responses and then return `X-Cashu` refund.
+3. Introduce a later/refund-claim mechanism, which would be a larger protocol change.
+
+## Summary
+
+- PPQ private models are billed per actual input/output tokens.
+- Private model rates are available from `GET /v1/models?type=all`.
+- Current Routstr EHBP billing at max cost is wrong for PPQ private models.
+- Direct Tinfoil integration inside Routstr would enable exact usage billing but would make Routstr see plaintext.
+- A blind EHBP relay preserves privacy but requires PPQ/Tinfoil to expose usage/cost in plaintext headers/trailers.
+- The preferred solution is to keep Routstr blind and have PPQ return billing metadata outside the encrypted body.
+
+## Implementation status
+
+Direct Tinfoil upstream integration is implemented in `routstr/upstream/tinfoil.py`
+and `routstr/upstream/ehbp.py`.
+
+### What was built
+
+- `TinfoilUpstreamProvider` (`provider_type = "tinfoil"`):
+  - Base URL: `https://inference.tinfoil.sh`
+  - Fetches models from the public `GET /v1/models` endpoint (no auth needed).
+  - Parses Tinfoil's pricing (`inputTokenPricePer1M`, `outputTokenPricePer1M`,
+    `requestPrice`) into the standard `Model`/`Pricing` schema.
+  - `supports_ehbp = True` — acts as a blind EHBP relay.
+  - `get_ehbp_forwarding_target()` returns a target that includes
+    `X-Tinfoil-Request-Usage-Metrics: true`.
+  - `forward_get_request()` proxies `/attestation` to `https://atc.tinfoil.sh/attestation`
+    so the SDK can fetch attestation bundles through Routstr.
+  - Registered in `routstr/upstream/__init__.py` and seeded from
+    `TINFOIL_API_KEY` env var.
+
+- `routstr/upstream/ehbp.py`:
+  - `parse_tinfoil_usage_metrics()` parses
+    `prompt=N,completion=N[,total=N][,model=<name>]` into an OpenAI-style
+    usage dict. The `model` field (added in tinfoilsh/confidential-model-router
+    PR #385) is extracted as a string.
+  - `_resolve_ehbp_target_url()` overrides the forwarding URL with
+    `X-Tinfoil-Enclave-Url` when the SDK sends it.
+  - `_strip_proxy_headers()` removes `X-Routstr-Model`,
+    `X-Tinfoil-Enclave-Url`, and `X-Tinfoil-Request-Usage-Metrics` before
+    forwarding to the enclave.
+  - `_compute_ehbp_actual_cost()` converts the usage header into msats via
+    `calculate_cost()`, clamped to `[min_request_msat, max_cost_for_model]`.
+    When the header's `model=<name>` differs from the requested model, the
+    actual served model's pricing is used for cost calculation.
+  - `forward_ehbp_request()` (bearer auth): if `X-Tinfoil-Usage-Metrics` is
+    present in the response header, finalizes with `adjust_payment_for_tokens()`
+    for exact billing; otherwise falls back to max-cost. Billing uses the
+    actual served model when it differs from the requested one.
+  - `forward_ehbp_x_cashu_request()`: if usage is available, computes the
+    refund from actual cost instead of max cost, using the actual served
+    model's pricing when applicable.
+
+- `routstr/proxy.py`: `/attestation` and `/tee/attestation` paths are forwarded
+  to Tinfoil upstreams without model/cost/auth lookups.
+
+### Billing behavior
+
+| Request shape | Usage source | Billing |
+|---|---|---|
+| Bearer, non-streaming | `X-Tinfoil-Usage-Metrics` response header | Exact token cost via `adjust_payment_for_tokens` |
+| Bearer, streaming | `X-Tinfoil-Usage-Metrics` HTTP trailer | Exact token cost (h11 captures trailers) |
+| Bearer, no usage header/trailer | N/A | Max-cost fallback |
+| X-Cashu, non-streaming | `X-Tinfoil-Usage-Metrics` response header | Refund = `redeemed - actual_cost` |
+| X-Cashu, streaming | `X-Tinfoil-Usage-Metrics` HTTP trailer | Refund = `redeemed - actual_cost` (h11 captures trailers) |
+| X-Cashu, no usage header/trailer | N/A | Refund = `redeemed - max_cost` |
+
+### Cost response headers
+
+Since EHBP response bodies are opaque encrypted blobs, per-request cost cannot
+be injected into the JSON body (as done in the normal proxy flow). Instead,
+Routstr returns cost info as response headers:
+
+| Header | Auth | Description |
+|---|---|---|
+| `X-Routstr-Cost-Msats` | Bearer, X-Cashu | Total msats charged for this request |
+| `X-Routstr-Cost-Usd` | Bearer | USD equivalent of the charge |
+| `X-Routstr-Input-Cost-Msats` | Bearer, X-Cashu | msats attributed to input tokens |
+| `X-Routstr-Output-Cost-Msats` | Bearer, X-Cashu | msats attributed to output tokens |
+
+The client/Tinfoil SDK can read these headers from the HTTP response without
+needing to decrypt the body.
+
+### Setup
+
+```bash
+TINFOIL_API_KEY=your-tinfoil-api-key
+```
+
+The provider is auto-seeded on first startup.
+
+### Usage metrics header format
+
+Tinfoil returns usage metrics in the `X-Tinfoil-Usage-Metrics` response header
+(non-streaming) or HTTP trailer (streaming) when `X-Tinfoil-Request-Usage-Metrics:
+true` is sent. As of tinfoilsh/confidential-model-router PR #385, the format is:
+
+```
+prompt=<prompt_tokens>,completion=<completion_tokens>,total=<total_tokens>,model=<served_model>
+```
+
+The `model` field carries the actual model name served by the enclave.
+Routstr uses this to:
+
+- Verify the served model matches the expected upstream model. The comparison
+  uses ``model_obj.forwarded_model_id`` (the actual upstream ID, e.g.
+  ``glm-5-2``) rather than ``model_obj.id`` (the client-facing alias, e.g.
+  ``tinfoil-glm-5-2``), so aliased models don't trigger a spurious mismatch.
+- When they genuinely differ (Tinfoil served a different upstream model than
+  expected), look up the actual served model's pricing and use it for billing.
+  The reverse lookup uses ``get_model_instance``, which resolves
+  ``forwarded_model_id`` values registered as routable aliases.
+- Log the discrepancy for observability.
+
+If the actual model is not found in Routstr's model registry, billing falls
+back to the requested model's pricing.
+
+### What still needs verification
+
+- ~~End-to-end test with a real Tinfoil SDK client against a Routstr node with
+  `TINFOIL_API_KEY` set.~~ Verified: both non-streaming (header) and streaming
+  (trailer) responses include `model=<name>`.
+- Streaming trailer capture is implemented by buffering the encrypted response
+  in `forward_with_trailer()` and then using the dedicated EHBP payment
+  finalizers for bearer and X-Cashu requests. This provides actual-cost billing
+  today, at the cost of full time-to-last-byte latency for streaming responses.
+- Whether Tinfoil's `/v1/responses` endpoint also returns usage metrics
+  headers or trailers.

migrations/versions/7f2843d3f4e4_add_reservation_release_idempotency_.py

@@ -0,0 +1,62 @@
+"""add reservation release idempotency records
+
+Revision ID: 7f2843d3f4e4
+Revises: fc4fa29630d2
+Create Date: 2026-07-24 02:06:06.066726
+"""
+
+from __future__ import annotations
+
+import sqlalchemy as sa
+from alembic import op
+
+revision = "7f2843d3f4e4"
+down_revision = "fc4fa29630d2"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    op.create_table(
+        "reservation_releases",
+        sa.Column("id", sa.String(), nullable=False),
+        sa.Column("key_hash", sa.String(), nullable=False),
+        sa.Column("billing_key_hash", sa.String(), nullable=False),
+        sa.Column("reserved_msats", sa.Integer(), nullable=False),
+        sa.Column(
+            "status", sa.String(), nullable=False, server_default="active"
+        ),
+        sa.Column("created_at", sa.Integer(), nullable=False),
+        sa.PrimaryKeyConstraint("id"),
+    )
+    op.create_index(
+        "ix_reservation_releases_key_hash",
+        "reservation_releases",
+        ["key_hash"],
+    )
+    op.create_index(
+        "ix_reservation_releases_billing_key_hash",
+        "reservation_releases",
+        ["billing_key_hash"],
+    )
+    op.create_index(
+        "ix_reservation_releases_status_created_at",
+        "reservation_releases",
+        ["status", "created_at"],
+    )
+
+
+def downgrade() -> None:
+    op.drop_index(
+        "ix_reservation_releases_status_created_at",
+        table_name="reservation_releases",
+    )
+    op.drop_index(
+        "ix_reservation_releases_billing_key_hash",
+        table_name="reservation_releases",
+    )
+    op.drop_index(
+        "ix_reservation_releases_key_hash",
+        table_name="reservation_releases",
+    )
+    op.drop_table("reservation_releases")

migrations/versions/9c4d8e2f1a6b_repair_fee_payout_checkpoint_columns.py

@@ -0,0 +1,47 @@
+"""repair missing fee payout checkpoint columns
+
+Revision ID: 9c4d8e2f1a6b
+Revises: 7f2843d3f4e4
+Create Date: 2026-07-25 00:00:00.000000
+"""
+
+from __future__ import annotations
+
+import sqlalchemy as sa
+from alembic import op
+
+revision = "9c4d8e2f1a6b"
+down_revision = "7f2843d3f4e4"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    """Repair databases stamped past the original checkpoint migration."""
+    conn = op.get_bind()
+    columns = {
+        column["name"] for column in sa.inspect(conn).get_columns("routstr_fees")
+    }
+
+    if "payout_in_progress_msats" not in columns:
+        op.add_column(
+            "routstr_fees",
+            sa.Column(
+                "payout_in_progress_msats",
+                sa.Integer(),
+                nullable=False,
+                server_default="0",
+            ),
+        )
+
+    if "payout_started_at" not in columns:
+        op.add_column(
+            "routstr_fees",
+            sa.Column("payout_started_at", sa.Integer(), nullable=True),
+        )
+
+
+def downgrade() -> None:
+    # The preceding revision already expects both columns. This migration only
+    # repairs schema drift, so downgrading it must preserve the expected schema.
+    pass

migrations/versions/aa50fde387a2_add_refund_sweep_claim_lease.py

@@ -0,0 +1,39 @@
+"""add refund sweep claim lease
+
+Revision ID: aa50fde387a2
+Revises: 9c4d8e2f1a6b
+Create Date: 2026-07-26 12:50:10.509217
+"""
+
+from __future__ import annotations
+
+import sqlalchemy as sa
+from alembic import op
+
+revision = "aa50fde387a2"
+down_revision = "9c4d8e2f1a6b"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    conn = op.get_bind()
+    columns = {
+        column["name"]
+        for column in sa.inspect(conn).get_columns("cashu_transactions")
+    }
+    if "sweep_started_at" not in columns:
+        op.add_column(
+            "cashu_transactions",
+            sa.Column("sweep_started_at", sa.Integer(), nullable=True),
+        )
+
+
+def downgrade() -> None:
+    conn = op.get_bind()
+    columns = {
+        column["name"]
+        for column in sa.inspect(conn).get_columns("cashu_transactions")
+    }
+    if "sweep_started_at" in columns:
+        op.drop_column("cashu_transactions", "sweep_started_at")

migrations/versions/c6d7e8f9a0b1_add_slug_to_upstream_providers.py

@@ -0,0 +1,88 @@
+"""add slug to upstream_providers
+
+Revision ID: c6d7e8f9a0b1
+Revises: b5e7c9d1f3a2
+Create Date: 2026-06-29 00:00:00.000000
+"""
+
+from __future__ import annotations
+
+import sqlalchemy as sa
+from alembic import op
+
+from routstr.core.provider_slugs import provider_slug_base, provider_slug_candidate
+
+revision = "c6d7e8f9a0b1"
+down_revision = "b5e7c9d1f3a2"
+branch_labels = None
+depends_on = None
+
+
+def _allocate_backfill_slug(provider_type: str, reserved_slugs: set[str]) -> str:
+    base = provider_slug_base(provider_type)
+    suffix_number = 1
+    while True:
+        candidate = provider_slug_candidate(base, suffix_number)
+        if candidate not in reserved_slugs:
+            reserved_slugs.add(candidate)
+            return candidate
+        suffix_number += 1
+
+
+def _backfill_provider_slugs(conn: sa.Connection) -> None:
+    existing_rows = conn.execute(
+        sa.text(
+            "SELECT slug FROM upstream_providers "
+            "WHERE slug IS NOT NULL AND slug != ''"
+        )
+    )
+    reserved_slugs = {str(row.slug).lower() for row in existing_rows}
+
+    rows_to_backfill = conn.execute(
+        sa.text(
+            "SELECT id, provider_type FROM upstream_providers "
+            "WHERE slug IS NULL OR slug = '' "
+            "ORDER BY id"
+        )
+    )
+    for row in rows_to_backfill:
+        slug = _allocate_backfill_slug(str(row.provider_type), reserved_slugs)
+        conn.execute(
+            sa.text("UPDATE upstream_providers SET slug = :slug WHERE id = :id"),
+            {"slug": slug, "id": row.id},
+        )
+
+
+def upgrade() -> None:
+    conn = op.get_bind()
+    inspector = sa.inspect(conn)
+    columns = {c["name"] for c in inspector.get_columns("upstream_providers")}
+
+    if "slug" not in columns:
+        op.add_column(
+            "upstream_providers",
+            sa.Column("slug", sa.String(), nullable=True),
+        )
+
+    _backfill_provider_slugs(conn)
+
+    existing_indexes = {idx["name"] for idx in inspector.get_indexes("upstream_providers")}
+    if "ix_upstream_providers_slug" not in existing_indexes:
+        op.create_index(
+            "ix_upstream_providers_slug",
+            "upstream_providers",
+            ["slug"],
+            unique=True,
+        )
+
+
+def downgrade() -> None:
+    conn = op.get_bind()
+    inspector = sa.inspect(conn)
+    existing_indexes = {idx["name"] for idx in inspector.get_indexes("upstream_providers")}
+    if "ix_upstream_providers_slug" in existing_indexes:
+        op.drop_index("ix_upstream_providers_slug", table_name="upstream_providers")
+
+    columns = {c["name"] for c in inspector.get_columns("upstream_providers")}
+    if "slug" in columns:
+        op.drop_column("upstream_providers", "slug")

migrations/versions/d7e8f9a0b1c2_add_fee_payout_checkpoint.py

@@ -0,0 +1,37 @@
+"""add fee payout checkpoint
+
+Revision ID: d7e8f9a0b1c2
+Revises: c6d7e8f9a0b1
+Create Date: 2026-07-18 00:00:00.000000
+"""
+
+from __future__ import annotations
+
+import sqlalchemy as sa
+from alembic import op
+
+revision = "d7e8f9a0b1c2"
+down_revision = "c6d7e8f9a0b1"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    op.add_column(
+        "routstr_fees",
+        sa.Column(
+            "payout_in_progress_msats",
+            sa.Integer(),
+            nullable=False,
+            server_default="0",
+        ),
+    )
+    op.add_column(
+        "routstr_fees",
+        sa.Column("payout_started_at", sa.Integer(), nullable=True),
+    )
+
+
+def downgrade() -> None:
+    op.drop_column("routstr_fees", "payout_started_at")
+    op.drop_column("routstr_fees", "payout_in_progress_msats")

migrations/versions/fc4fa29630d2_add_secrets_table.py

@@ -0,0 +1,51 @@
+"""add secrets table
+
+Revision ID: fc4fa29630d2
+Revises: d7e8f9a0b1c2
+Create Date: 2026-07-23 00:00:00.000000
+
+Creates the node-level singleton secret store (issue #553). Schema only; moving
+any legacy plaintext into the encrypted/hashed columns happens at bootstrap,
+where the live ROUTSTR_SECRET_KEY is available. ``nsec_state`` records the vault's
+ownership of the nsec (legacy | encrypted | cleared), so a cleared identity is
+never resurrected from a stale legacy ``NSEC`` env var / settings blob on the next
+boot.
+"""
+
+import sqlalchemy as sa
+import sqlmodel
+from alembic import op
+
+revision = "fc4fa29630d2"
+down_revision = "d7e8f9a0b1c2"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    op.create_table(
+        "secrets",
+        sa.Column("id", sa.Integer(), nullable=False),
+        sa.Column(
+            "admin_password_hash",
+            sqlmodel.sql.sqltypes.AutoString(),
+            nullable=True,
+        ),
+        sa.Column(
+            "encrypted_nsec",
+            sqlmodel.sql.sqltypes.AutoString(),
+            nullable=True,
+        ),
+        sa.Column(
+            "nsec_state",
+            sqlmodel.sql.sqltypes.AutoString(),
+            nullable=False,
+            server_default="legacy",
+        ),
+        sa.Column("updated_at", sa.Integer(), nullable=True),
+        sa.PrimaryKeyConstraint("id"),
+    )
+
+
+def downgrade() -> None:
+    op.drop_table("secrets")

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "routstr"
-version = "0.4.3"
+version = "0.4.4"
 description = "Payment proxy for your LLM endpoint using cashu and nostr."
 readme = "README.md"
 requires-python = ">=3.11"
@@ -10,6 +10,7 @@ dependencies = [
     "aiosqlite>=0.20",
     "sqlmodel>=0.0.24",
     "httpx[socks]>=0.25.2",
+    "h11>=0.14",
     "greenlet>=3.2.1",
     "alembic>=1.13",
     "python-json-logger>=2.0.0",

routstr/algorithm.py

@@ -86,10 +86,12 @@ def get_provider_penalty(provider: "BaseUpstreamProvider") -> float:
 
 def create_model_mappings(
     upstreams: list["BaseUpstreamProvider"],
-    overrides_by_id: dict[str, tuple],
-    disabled_model_ids: set[str],
+    overrides_by_key: dict[tuple[str, int], tuple],
+    disabled_model_keys: set[tuple[str, int]],
 ) -> tuple[
-    dict[str, "Model"], dict[str, list["BaseUpstreamProvider"]], dict[str, "Model"]
+    dict[str, "Model"],
+    dict[str, list[tuple["Model", "BaseUpstreamProvider"]]],
+    dict[str, "Model"],
 ]:
     """Create optimal model mappings based on cost and provider preferences.
 
@@ -97,7 +99,9 @@ def create_model_mappings(
     and creates three mappings based on cost optimization:
 
     1. model_instances: alias -> Model (all model aliases mapped to their Model objects)
-    2. provider_map: alias -> List[UpstreamProvider] (sorted list of providers for each alias)
+    2. provider_map: alias -> List[(Model, UpstreamProvider)] (sorted candidate
+       list for each alias; each provider is paired with ITS OWN model so
+       failover can forward and bill the candidate that actually serves)
     3. unique_models: base_id -> Model (unique models without provider prefixes)
 
     The algorithm:
@@ -107,8 +111,9 @@ def create_model_mappings(
 
     Args:
         upstreams: List of all upstream provider instances
-        overrides_by_id: Dict of model overrides from database {model_id: (ModelRow, fee)}
-        disabled_model_ids: Set of model IDs that should be excluded
+        overrides_by_key: Dict of model overrides from database
+            {(model_id_lower, upstream_provider_id): (ModelRow, fee)}
+        disabled_model_keys: Set of provider-scoped model keys that should be excluded
 
     Returns:
         Tuple of (model_instances, provider_map, unique_models)
@@ -179,14 +184,22 @@ def create_model_mappings(
         """Process all models from a given provider."""
         upstream_prefix = getattr(upstream, "upstream_name", None)
         provider_key = get_provider_identity(upstream)
+        upstream_db_id = getattr(upstream, "db_id", None)
 
         for model in upstream.get_cached_models():
-            if not model.enabled or model.id in disabled_model_ids:
+            model_key = (
+                (model.id.lower(), upstream_db_id)
+                if isinstance(upstream_db_id, int)
+                else None
+            )
+            if not model.enabled or (
+                model_key is not None and model_key in disabled_model_keys
+            ):
                 continue
 
-            # Apply overrides if present
-            if model.id in overrides_by_id:
-                override_row, provider_fee = overrides_by_id[model.id]
+            # Apply overrides only for this provider's model row.
+            if model_key is not None and model_key in overrides_by_key:
+                override_row, provider_fee = overrides_by_key[model_key]
                 model_to_use = _row_to_model(
                     override_row, apply_provider_fee=True, provider_fee=provider_fee
                 )
@@ -237,13 +250,10 @@ def create_model_mappings(
 
     # Include enabled DB overrides even when provider discovery misses models.
     # This is important for deployment-based providers like Azure.
-    for model_id, override_data in overrides_by_id.items():
-        if model_id in disabled_model_ids:
+    for (model_id, upstream_provider_id), override_data in overrides_by_key.items():
+        if (model_id, upstream_provider_id) in disabled_model_keys:
             continue
         override_row, provider_fee = override_data
-        upstream_provider_id = getattr(override_row, "upstream_provider_id", None)
-        if not isinstance(upstream_provider_id, int):
-            continue
 
         upstream_for_override = providers_by_db_id.get(upstream_provider_id)
         if upstream_for_override is None:
@@ -321,7 +331,7 @@ def create_model_mappings(
 
     # Sort candidates and build final maps
     model_instances: dict[str, "Model"] = {}
-    provider_map: dict[str, list["BaseUpstreamProvider"]] = {}
+    provider_map: dict[str, list[tuple["Model", "BaseUpstreamProvider"]]] = {}
 
     def alias_priority(model: "Model", alias: str) -> int:
         """Rank how strong the mapping of alias->model is.
@@ -368,13 +378,13 @@ def create_model_mappings(
 
         best_model, best_provider = items[0]
         model_instances[alias] = best_model
-        provider_map[alias] = [p for _, p in items]
+        provider_map[alias] = list(items)
 
     # Log provider distribution (using top provider for stats)
     provider_counts: dict[str, int] = {}
-    for providers in provider_map.values():
-        if providers:
-            provider = providers[0]
+    for candidate_list in provider_map.values():
+        if candidate_list:
+            provider = candidate_list[0][1]
             provider_name = getattr(provider, "upstream_name", "unknown")
             provider_counts[provider_name] = provider_counts.get(provider_name, 0) + 1
 

routstr/auth.py

@@ -3,16 +3,25 @@ import hashlib
 import math
 import random
 import time
+import uuid
+from contextvars import ContextVar
+from dataclasses import dataclass
 from datetime import datetime
-from typing import Optional
+from typing import TYPE_CHECKING, Optional
 
 from fastapi import HTTPException
-from sqlalchemy import case
+from sqlalchemy import case, inspect
 from sqlalchemy.exc import IntegrityError
 from sqlmodel import col, select, update
 
 from .core import get_logger
-from .core.db import ApiKey, AsyncSession, accumulate_routstr_fee
+from .core.db import (
+    ApiKey,
+    AsyncSession,
+    ReservationRelease,
+    accumulate_routstr_fee,
+    create_session,
+)
 from .core.settings import settings
 from .payment.cost_calculation import (
     CostData,
@@ -20,17 +29,47 @@ from .payment.cost_calculation import (
     MaxCostData,
     calculate_cost,
 )
-from .wallet import credit_balance, deserialize_token_from_string
+from .wallet import (
+    classify_redemption_error,
+    credit_balance,
+    deserialize_token_from_string,
+    wallet_operation_guard,
+)
+
+if TYPE_CHECKING:
+    from .payment.models import Model
 
 logger = get_logger(__name__)
 payments_logger = get_logger("routstr.payments")
 
 # Routstr platform fee constants
 ROUTSTR_FEE_PERCENT: float = 2.1
-ROUTSTR_LN_ADDRESS: str = "npub130mznv74rxs032peqym6g3wqavh472623mt3z5w73xq9r6qqdufs7ql29s@npub.cash"
+ROUTSTR_LN_ADDRESS: str = (
+    "npub130mznv74rxs032peqym6g3wqavh472623mt3z5w73xq9r6qqdufs7ql29s@npub.cash"
+)
 ROUTSTR_FEE_PAYOUT_INTERVAL_SECONDS: int = 900
 ROUTSTR_FEE_DEFAULT_PAYOUT: int = 200
 
+
+@dataclass(frozen=True)
+class ReservationSnapshot:
+    release_id: str
+    key_hash: str
+    billing_key_hash: str
+    reserved_msats: int
+
+
+_current_reservation: ContextVar[ReservationSnapshot | None] = ContextVar(
+    "current_billing_reservation", default=None
+)
+
+
+def _clear_current_reservation(snapshot: ReservationSnapshot) -> None:
+    current = _current_reservation.get()
+    if current is not None and current.release_id == snapshot.release_id:
+        _current_reservation.set(None)
+
+
 # TODO: implement prepaid api key (not like it was before)
 # PREPAID_API_KEY = os.environ.get("PREPAID_API_KEY", None)
 # PREPAID_BALANCE = int(os.environ.get("PREPAID_BALANCE", "0")) * 1000  # Convert to msats
@@ -78,12 +117,66 @@ async def check_and_reset_limit(key: ApiKey, session: AsyncSession) -> bool:
     return False
 
 
+def redemption_error_to_http_exception(error: Exception) -> HTTPException:
+    """Map a Cashu token redemption failure to a sanitized client-facing error.
+
+    Thin wrapper over the shared :func:`classify_redemption_error` so the bearer
+    path stays identical to the X-Cashu and top-up paths.
+    """
+    classified = classify_redemption_error(error)
+    if classified is None:
+        return HTTPException(
+            status_code=500,
+            detail={
+                "error": {
+                    "message": "Internal error during token redemption",
+                    "type": "api_error",
+                    "code": "internal_error",
+                }
+            },
+        )
+    error_type, status_code, message, error_code = classified
+    return HTTPException(
+        status_code=status_code,
+        detail={
+            "error": {
+                "message": message,
+                "type": error_type,
+                "code": error_code,
+            }
+        },
+    )
+
+
 async def validate_bearer_key(
     bearer_key: str,
     session: AsyncSession,
     refund_address: Optional[str] = None,
     key_expiry_time: Optional[int] = None,
     min_cost: int = 0,
+) -> ApiKey:
+    if bearer_key.startswith("cashu"):
+        # Acquire before the first lookup/flush so concurrent token creation
+        # cannot hold SQLite write transactions while waiting to mutate proofs.
+        async with wallet_operation_guard():
+            return await _validate_bearer_key_locked(
+                bearer_key,
+                session,
+                refund_address,
+                key_expiry_time,
+                min_cost,
+            )
+    return await _validate_bearer_key_locked(
+        bearer_key, session, refund_address, key_expiry_time, min_cost
+    )
+
+
+async def _validate_bearer_key_locked(
+    bearer_key: str,
+    session: AsyncSession,
+    refund_address: Optional[str] = None,
+    key_expiry_time: Optional[int] = None,
+    min_cost: int = 0,
 ) -> ApiKey:
     """
     Validates the provided API key using SQLModel.
@@ -216,7 +309,17 @@ async def validate_bearer_key(
 
         try:
             hashed_key = hashlib.sha256(bearer_key.encode()).hexdigest()
-            token_obj = deserialize_token_from_string(bearer_key)
+            try:
+                token_obj = deserialize_token_from_string(bearer_key)
+            except Exception as decode_error:
+                # A malformed token is a bad token (400 invalid_cashu_token via
+                # the shared taxonomy), not an auth failure (401) — otherwise it
+                # would fall through to the generic "Invalid API key" handler.
+                raise redemption_error_to_http_exception(
+                    ValueError(
+                        f"Invalid Cashu token: could not decode token ({decode_error})"
+                    )
+                ) from decode_error
             logger.debug(
                 "Generated token hash", extra={"hash_preview": hashed_key[:16] + "..."}
             )
@@ -334,19 +437,32 @@ async def validate_bearer_key(
                         "error_type": type(credit_error).__name__,
                     },
                 )
-                raise credit_error
+                await session.rollback()
+                raise redemption_error_to_http_exception(credit_error) from credit_error
 
             if msats <= 0:
                 logger.error(
                     "Token redemption returned zero or negative amount",
                     extra={"msats": msats, "key_hash": hashed_key[:8] + "..."},
                 )
-                # Defense-in-depth: credit_balance now refuses to commit on a
-                # zero/negative redemption, but if a row was nonetheless
-                # persisted, drop it so we never leave an orphan zero-balance key.
+                # Defense-in-depth: credit_balance already raises
+                # ValueError("Redeemed token amount must be positive…") before
+                # returning (wallet.py), so this branch is only reachable if a
+                # zero/negative row was somehow persisted; drop it so we never
+                # leave an orphan zero-balance key. Reuse the shared taxonomy
+                # (cashu_error) so the envelope matches the mapper above.
                 await session.delete(new_key)
                 await session.commit()
-                raise Exception("Token redemption failed")
+                raise HTTPException(
+                    status_code=400,
+                    detail={
+                        "error": {
+                            "message": "Failed to redeem Cashu token: token yielded no value",
+                            "type": "cashu_error",
+                            "code": "cashu_token_zero_value",
+                        }
+                    },
+                )
 
             await session.refresh(new_key)
             await session.commit()
@@ -364,6 +480,7 @@ async def validate_bearer_key(
         except HTTPException:
             raise
         except Exception as e:
+            await session.rollback()
             logger.error(
                 "Cashu token redemption failed",
                 extra={
@@ -378,7 +495,7 @@ async def validate_bearer_key(
                 status_code=401,
                 detail={
                     "error": {
-                        "message": f"Invalid or expired Cashu key: {str(e)}",
+                        "message": "Invalid or expired Cashu key",
                         "type": "invalid_request_error",
                         "code": "invalid_api_key",
                     }
@@ -533,6 +650,16 @@ async def pay_for_request(
         },
     )
 
+    # Create the durable reservation identity before changing aggregate balances.
+    # The row and balance updates commit together, so every reserved amount has one
+    # owner that can reach exactly one terminal state.
+    reservation = ReservationSnapshot(
+        release_id=uuid.uuid4().hex,
+        key_hash=key.hashed_key,
+        billing_key_hash=billing_key.hashed_key,
+        reserved_msats=cost_per_request,
+    )
+
     # Charge the base cost for the request atomically to avoid race conditions
     reserved_at_now = int(time.time())
     stmt = (
@@ -594,22 +721,83 @@ async def pay_for_request(
         child_result = await session.exec(child_stmt)  # type: ignore[call-overload]
 
         if child_result.rowcount == 0:
+            # Build the error before rollback expires ORM attributes.
+            limit_message = (
+                f"Balance limit exceeded: {key.balance_limit} mSats limit. "
+                f"{key.total_spent} already spent ({key.reserved_balance} reserved), "
+                f"{cost_per_request} required for this request."
+            )
+            # The parent reservation update already ran in this transaction.
+            # Roll it back before failover code attempts to restore the previous
+            # reservation; otherwise that later commit can persist both updates.
+            await session.rollback()
             raise HTTPException(
                 status_code=402,
                 detail={
                     "error": {
-                        "message": f"Balance limit exceeded: {key.balance_limit} mSats limit. {key.total_spent} already spent ({key.reserved_balance} reserved), {cost_per_request} required for this request.",
+                        "message": limit_message,
                         "type": "insufficient_quota",
                         "code": "balance_limit_exceeded",
                     }
                 },
             )
 
-    await session.commit()
+    session.add(
+        ReservationRelease(
+            id=reservation.release_id,
+            key_hash=reservation.key_hash,
+            billing_key_hash=reservation.billing_key_hash,
+            reserved_msats=reservation.reserved_msats,
+            status="active",
+        )
+    )
+    # Publish the identity before commit. If the commit succeeds but its
+    # acknowledgement is interrupted, exact cleanup can still recover the
+    # durable row. A definitely failed commit is harmless because every
+    # terminal transition validates that row before touching balances.
+    _current_reservation.set(reservation)
+    try:
+        await session.commit()
+    except BaseException:
+        # The database may have committed even if acknowledgement was cancelled
+        # or the connection failed. Reconcile using a fresh transaction and the
+        # exact durable identity; no upstream request has started yet.
+        try:
+            await session.rollback()
+        except Exception:
+            pass
+        try:
+            async with create_session() as cleanup_session:
+                record = await cleanup_session.get(
+                    ReservationRelease, reservation.release_id
+                )
+                if record is not None and record.status == "active":
+                    await _transition_reservation_to_released(
+                        reservation,
+                        cleanup_session,
+                        decrement_requests=True,
+                        idempotent_success=True,
+                    )
+        except Exception:
+            logger.exception(
+                "Failed to reconcile ambiguous reservation commit",
+                extra={"reservation_id": reservation.release_id},
+            )
+        finally:
+            _clear_current_reservation(reservation)
+        raise
 
-    await session.refresh(billing_key)
-    if billing_key.hashed_key != key.hashed_key:
-        await session.refresh(key)
+    try:
+        await session.refresh(billing_key)
+        if billing_key.hashed_key != key.hashed_key:
+            await session.refresh(key)
+    except Exception:
+        # The reservation transaction is already committed and durable. Logging
+        # refresh failures must not make the caller treat it as unreserved.
+        logger.exception(
+            "Reservation committed but post-commit refresh failed",
+            extra={"reservation_id": reservation.release_id},
+        )
 
     logger.info(
         "Payment processed successfully",
@@ -639,83 +827,206 @@ async def pay_for_request(
 
 
 async def revert_pay_for_request(
-    key: ApiKey, session: AsyncSession, cost_per_request: int
+    key: ApiKey,
+    session: AsyncSession,
+    cost_per_request: int,
+    reservation_snapshot: ReservationSnapshot | None = None,
 ) -> bool:
-    """Revert a previously reserved payment. Returns True if revert succeeded,
-    False if the reservation was already released (prevents negative reserved_balance)."""
-    billing_key = await get_billing_key(key, session)
-
-    # Keep reserved_at while other reservations remain
-    cleared_reserved_at = case(
-        (col(ApiKey.reserved_balance) - cost_per_request > 0, col(ApiKey.reserved_at)),
-        else_=None,
+    """Revert the current request's durable reservation exactly once."""
+    snapshot = reservation_snapshot or await get_reservation_snapshot(key, session)
+    await _validate_reservation_snapshot(key, snapshot, session, require_active=False)
+    if cost_per_request != snapshot.reserved_msats:
+        return False
+    return await _transition_reservation_to_released(
+        snapshot,
+        session,
+        decrement_requests=True,
+        idempotent_success=False,
     )
 
-    stmt = (
+
+async def _validate_reservation_snapshot(
+    key: ApiKey,
+    snapshot: ReservationSnapshot,
+    session: AsyncSession,
+    *,
+    require_active: bool = True,
+) -> None:
+    """Reject cross-request or forged reservation handles before any mutation."""
+    state = inspect(key)
+    identity = state.identity if state is not None else None
+    key_hash = str(identity[0]) if identity else key.__dict__.get("hashed_key")
+    if snapshot.key_hash != key_hash:
+        raise RuntimeError("Billing reservation does not belong to this key")
+
+    persisted_key = await session.get(ApiKey, snapshot.key_hash)
+    if persisted_key is None:
+        raise RuntimeError("Billing reservation key no longer exists")
+    expected_billing_hash = persisted_key.parent_key_hash or persisted_key.hashed_key
+    if snapshot.billing_key_hash != expected_billing_hash:
+        raise RuntimeError("Billing reservation does not belong to this billing key")
+
+    record = await session.get(ReservationRelease, snapshot.release_id)
+    if (
+        record is None
+        or (require_active and record.status != "active")
+        or record.key_hash != snapshot.key_hash
+        or record.billing_key_hash != snapshot.billing_key_hash
+        or record.reserved_msats != snapshot.reserved_msats
+    ):
+        raise RuntimeError("Billing reservation record does not match the request")
+
+
+async def get_reservation_snapshot(
+    key: ApiKey, session: AsyncSession
+) -> ReservationSnapshot:
+    """Return the durable reservation created for the current request."""
+    snapshot = _current_reservation.get()
+    if snapshot is None:
+        raise RuntimeError("No billing reservation is associated with this request")
+    await _validate_reservation_snapshot(key, snapshot, session)
+    return snapshot
+
+
+async def _transition_reservation_to_released(
+    snapshot: ReservationSnapshot,
+    session: AsyncSession,
+    *,
+    decrement_requests: bool,
+    idempotent_success: bool,
+) -> bool:
+    transition = (
+        update(ReservationRelease)
+        .where(col(ReservationRelease.id) == snapshot.release_id)
+        .where(col(ReservationRelease.status) == "active")
+        .where(col(ReservationRelease.key_hash) == snapshot.key_hash)
+        .where(col(ReservationRelease.billing_key_hash) == snapshot.billing_key_hash)
+        .where(col(ReservationRelease.reserved_msats) == snapshot.reserved_msats)
+        .values(status="released")
+    )
+    transition_result = await session.exec(transition)  # type: ignore[call-overload]
+    if transition_result.rowcount != 1:
+        await session.rollback()
+        existing = await session.get(ReservationRelease, snapshot.release_id)
+        return bool(
+            idempotent_success
+            and existing is not None
+            and existing.status == "released"
+            and existing.key_hash == snapshot.key_hash
+            and existing.billing_key_hash == snapshot.billing_key_hash
+            and existing.reserved_msats == snapshot.reserved_msats
+        )
+
+    values: dict[str, object] = {
+        "reserved_balance": col(ApiKey.reserved_balance) - snapshot.reserved_msats,
+        "reserved_at": case(
+            (
+                col(ApiKey.reserved_balance) - snapshot.reserved_msats > 0,
+                col(ApiKey.reserved_at),
+            ),
+            else_=None,
+        ),
+    }
+    if decrement_requests:
+        values["total_requests"] = col(ApiKey.total_requests) - 1
+
+    release_stmt = (
         update(ApiKey)
-        .where(col(ApiKey.hashed_key) == billing_key.hashed_key)
-        .where(col(ApiKey.reserved_balance) >= cost_per_request)
-        .values(
-            reserved_balance=col(ApiKey.reserved_balance) - cost_per_request,
-            reserved_at=cleared_reserved_at,
-            total_requests=col(ApiKey.total_requests) - 1,
-        )
+        .where(col(ApiKey.hashed_key) == snapshot.billing_key_hash)
+        .where(col(ApiKey.reserved_balance) >= snapshot.reserved_msats)
+        .values(**values)
     )
+    result = await session.exec(release_stmt)  # type: ignore[call-overload]
+    if result.rowcount != 1:
+        await session.rollback()
+        return False
 
-    result = await session.exec(stmt)  # type: ignore[call-overload]
-
-    # Also decrement total_requests and reserved_balance on the child key if it's different
-    if billing_key.hashed_key != key.hashed_key:
-        child_stmt = (
+    if snapshot.billing_key_hash != snapshot.key_hash:
+        child_release_stmt = (
             update(ApiKey)
-            .where(col(ApiKey.hashed_key) == key.hashed_key)
-            .where(col(ApiKey.reserved_balance) >= cost_per_request)
-            .values(
-                total_requests=col(ApiKey.total_requests) - 1,
-                reserved_balance=col(ApiKey.reserved_balance) - cost_per_request,
-                reserved_at=cleared_reserved_at,
-            )
+            .where(col(ApiKey.hashed_key) == snapshot.key_hash)
+            .where(col(ApiKey.reserved_balance) >= snapshot.reserved_msats)
+            .values(**values)
         )
-        await session.exec(child_stmt)  # type: ignore[call-overload]
+        child_result = await session.exec(  # type: ignore[call-overload]
+            child_release_stmt
+        )
+        if child_result.rowcount != 1:
+            await session.rollback()
+            return False
 
     await session.commit()
-    if result.rowcount == 0:
-        logger.warning(
-            "Revert skipped - reservation already released (no-op to prevent negative reserved_balance)",
-            extra={
-                "key_hash": key.hashed_key[:8] + "...",
-                "billing_key_hash": billing_key.hashed_key[:8] + "...",
-                "cost_to_revert": cost_per_request,
-                "current_reserved_balance": billing_key.reserved_balance,
-            },
-        )
-        return False
-    await session.refresh(billing_key)
-    if billing_key.hashed_key != key.hashed_key:
-        await session.refresh(key)
-    payments_logger.info(
-        "REVERT",
-        extra={
-            "event": "revert",
-            "key_hash": key.hashed_key[:8] + "...",
-            "billing_key_hash": billing_key.hashed_key[:8] + "...",
-            "cost_reverted": cost_per_request,
-            "balance": billing_key.balance,
-            "reserved_balance": billing_key.reserved_balance,
-        },
-    )
+    _clear_current_reservation(snapshot)
     return True
 
 
+async def release_reservation(
+    snapshot: ReservationSnapshot,
+    session: AsyncSession,
+    reserved_msats: int,
+) -> bool:
+    """Release one durable reservation exactly once without charging."""
+    if reserved_msats <= 0 or reserved_msats != snapshot.reserved_msats:
+        return False
+    return await _transition_reservation_to_released(
+        snapshot,
+        session,
+        decrement_requests=False,
+        idempotent_success=True,
+    )
+
+
+async def _claim_reservation_for_charge(
+    snapshot: ReservationSnapshot, session: AsyncSession
+) -> bool:
+    """Claim an active reservation in the caller's charge transaction."""
+    statement = (
+        update(ReservationRelease)
+        .where(col(ReservationRelease.id) == snapshot.release_id)
+        .where(col(ReservationRelease.status) == "active")
+        .where(col(ReservationRelease.key_hash) == snapshot.key_hash)
+        .where(col(ReservationRelease.billing_key_hash) == snapshot.billing_key_hash)
+        .where(col(ReservationRelease.reserved_msats) == snapshot.reserved_msats)
+        .values(status="charged")
+    )
+    result = await session.exec(statement)  # type: ignore[call-overload]
+    if result.rowcount == 1:
+        _clear_current_reservation(snapshot)
+        return True
+
+    await session.rollback()
+    return False
+
+
 async def adjust_payment_for_tokens(
-    key: ApiKey, response_data: dict, session: AsyncSession, deducted_max_cost: int
+    key: ApiKey,
+    response_data: dict,
+    session: AsyncSession,
+    deducted_max_cost: int,
+    model_obj: "Model | None" = None,
+    provider_fee: float | None = None,
+    reservation_snapshot: ReservationSnapshot | None = None,
 ) -> dict:
     """
     Adjusts the payment based on token usage in the response.
     This is called after the initial payment and the upstream request is complete.
     Returns cost data to be included in the response.
+
+    ``model_obj`` is the model that actually served the request; it is passed
+    through to ``calculate_cost`` so billing uses the serving candidate's
+    pricing instead of re-deriving it from the response's model string.
+
+    The response's usage object is normalized with the default union parser in
+    ``calculate_cost``.
     """
     billing_key = await get_billing_key(key, session)
+    reservation = reservation_snapshot or await get_reservation_snapshot(key, session)
+    await _validate_reservation_snapshot(
+        key, reservation, session, require_active=False
+    )
+    # The persisted amount is authoritative if request-level minimum pricing
+    # changed the caller's original estimate.
+    deducted_max_cost = reservation.reserved_msats
     model = response_data.get("model", "unknown")
 
     logger.debug(
@@ -731,50 +1042,21 @@ async def adjust_payment_for_tokens(
     )
 
     async def release_reservation_only() -> None:
-        """Fallback to release reservation without charging when main update fails."""
+        """Fallback to release this request's reservation without charging."""
         try:
-            release_stmt = (
-                update(ApiKey)
-                .where(col(ApiKey.hashed_key) == billing_key.hashed_key)
-                .where(col(ApiKey.reserved_balance) >= deducted_max_cost)
-                .values(
-                    reserved_balance=col(ApiKey.reserved_balance) - deducted_max_cost
-                )
+            released = await release_reservation(
+                reservation, session, reservation.reserved_msats
+            )
+            logger.warning(
+                "Released reservation without charging (fallback)"
+                if released
+                else "Reservation was already finalized; fallback skipped",
+                extra={
+                    "key_hash": key.hashed_key[:8] + "...",
+                    "billing_key_hash": billing_key.hashed_key[:8] + "...",
+                    "deducted_max_cost": deducted_max_cost,
+                },
             )
-            result = await session.exec(release_stmt)  # type: ignore[call-overload]
-
-            # Also release on child key if it's different
-            if billing_key.hashed_key != key.hashed_key:
-                child_release_stmt = (
-                    update(ApiKey)
-                    .where(col(ApiKey.hashed_key) == key.hashed_key)
-                    .where(col(ApiKey.reserved_balance) >= deducted_max_cost)
-                    .values(
-                        reserved_balance=col(ApiKey.reserved_balance)
-                        - deducted_max_cost
-                    )
-                )
-                await session.exec(child_release_stmt)  # type: ignore[call-overload]
-
-            await session.commit()
-            if result.rowcount == 0:  # type: ignore[union-attr]
-                logger.warning(
-                    "Release reservation skipped - already released (no-op to prevent negative reserved_balance)",
-                    extra={
-                        "key_hash": key.hashed_key[:8] + "...",
-                        "billing_key_hash": billing_key.hashed_key[:8] + "...",
-                        "deducted_max_cost": deducted_max_cost,
-                    },
-                )
-            else:
-                logger.warning(
-                    "Released reservation without charging (fallback)",
-                    extra={
-                        "key_hash": key.hashed_key[:8] + "...",
-                        "billing_key_hash": billing_key.hashed_key[:8] + "...",
-                        "deducted_max_cost": deducted_max_cost,
-                    },
-                )
         except Exception as e:
             logger.error(
                 "Failed to release reservation in fallback",
@@ -796,7 +1078,17 @@ async def adjust_payment_for_tokens(
                     extra={"error": str(e), "fee_msats": fee_msats},
                 )
 
-    match await calculate_cost(response_data, deducted_max_cost, session):
+    calculated_cost = await calculate_cost(
+        response_data, deducted_max_cost, model_obj, provider_fee
+    )
+    if not isinstance(calculated_cost, CostDataError):
+        if not await _claim_reservation_for_charge(reservation, session):
+            # A prior charge or release already owns this reservation. Returning
+            # the calculated metadata is safe; the aggregate balances must not
+            # be modified a second time.
+            return calculated_cost.dict()
+
+    match calculated_cost:
         case MaxCostData() as cost:
             logger.debug(
                 "Using max cost data (no token adjustment)",
@@ -824,8 +1116,10 @@ async def adjust_payment_for_tokens(
                 )
 
             safe_reserved = case(
-                (col(ApiKey.reserved_balance) >= deducted_max_cost,
-                 col(ApiKey.reserved_balance) - deducted_max_cost),
+                (
+                    col(ApiKey.reserved_balance) >= deducted_max_cost,
+                    col(ApiKey.reserved_balance) - deducted_max_cost,
+                ),
                 else_=0,
             )
 
@@ -843,8 +1137,10 @@ async def adjust_payment_for_tokens(
             # Also update total_spent and reserved_balance on the child key if it's different
             if billing_key.hashed_key != key.hashed_key:
                 child_safe_reserved = case(
-                    (col(ApiKey.reserved_balance) >= deducted_max_cost,
-                     col(ApiKey.reserved_balance) - deducted_max_cost),
+                    (
+                        col(ApiKey.reserved_balance) >= deducted_max_cost,
+                        col(ApiKey.reserved_balance) - deducted_max_cost,
+                    ),
                     else_=0,
                 )
                 child_stmt = (
@@ -954,8 +1250,10 @@ async def adjust_payment_for_tokens(
                     )
 
                 exact_safe_reserved = case(
-                    (col(ApiKey.reserved_balance) >= deducted_max_cost,
-                     col(ApiKey.reserved_balance) - deducted_max_cost),
+                    (
+                        col(ApiKey.reserved_balance) >= deducted_max_cost,
+                        col(ApiKey.reserved_balance) - deducted_max_cost,
+                    ),
                     else_=0,
                 )
 
@@ -973,8 +1271,10 @@ async def adjust_payment_for_tokens(
                 # Also update total_spent and reserved_balance on the child key if it's different
                 if billing_key.hashed_key != key.hashed_key:
                     child_exact_safe_reserved = case(
-                        (col(ApiKey.reserved_balance) >= deducted_max_cost,
-                         col(ApiKey.reserved_balance) - deducted_max_cost),
+                        (
+                            col(ApiKey.reserved_balance) >= deducted_max_cost,
+                            col(ApiKey.reserved_balance) - deducted_max_cost,
+                        ),
                         else_=0,
                     )
                     child_stmt = (
@@ -1013,85 +1313,91 @@ async def adjust_payment_for_tokens(
 
             # actual cost exceeded discounted reservation (due to tolerance_percentage)
             if cost_difference > 0:
-                # Always release the reservation and charge min(actual_cost, balance).
-                # Using a CASE expression makes this a single atomic UPDATE — no
-                # multi-level fallback needed and balance can never go negative.
-                chargeable = case(
-                    (col(ApiKey.balance) >= total_cost_msats, total_cost_msats),
-                    else_=col(ApiKey.balance),
-                )
-
-                finalize_stmt = (
-                    update(ApiKey)
-                    .where(col(ApiKey.hashed_key) == billing_key.hashed_key)
-                    .where(col(ApiKey.reserved_balance) >= deducted_max_cost)
-                    .values(
-                        reserved_balance=col(ApiKey.reserved_balance) - deducted_max_cost,
-                        balance=col(ApiKey.balance) - chargeable,
-                        total_spent=col(ApiKey.total_spent) + chargeable,
+                # Lock the billing row so the parent and child record the same
+                # database-determined charge under concurrent finalizations.
+                actual_charge_msats = 0
+                for attempt in range(5):
+                    locked_billing_key = (
+                        await session.exec(
+                            select(ApiKey)
+                            .where(col(ApiKey.hashed_key) == billing_key.hashed_key)
+                            .with_for_update()
+                            .execution_options(populate_existing=True)
+                        )
+                    ).one()
+                    observed_balance = locked_billing_key.balance
+                    actual_charge_msats = min(observed_balance, total_cost_msats)
+                    overrun_safe_reserved = case(
+                        (
+                            col(ApiKey.reserved_balance) >= deducted_max_cost,
+                            col(ApiKey.reserved_balance) - deducted_max_cost,
+                        ),
+                        else_=0,
                     )
-                )
-                result = await session.exec(finalize_stmt)  # type: ignore[call-overload]
+                    finalize_result = await session.exec(  # type: ignore[call-overload]
+                        update(ApiKey)
+                        .where(col(ApiKey.hashed_key) == billing_key.hashed_key)
+                        .where(col(ApiKey.balance) == observed_balance)
+                        .values(
+                            reserved_balance=overrun_safe_reserved,
+                            balance=col(ApiKey.balance) - actual_charge_msats,
+                            total_spent=col(ApiKey.total_spent) + actual_charge_msats,
+                        )
+                    )
+                    if finalize_result.rowcount == 1:
+                        break
+                    await session.rollback()
+                    if not await _claim_reservation_for_charge(reservation, session):
+                        return cost.dict()
+                else:
+                    await session.rollback()
+                    raise RuntimeError("Could not atomically finalize cost overrun")
 
                 if billing_key.hashed_key != key.hashed_key:
                     child_stmt = (
                         update(ApiKey)
                         .where(col(ApiKey.hashed_key) == key.hashed_key)
-                        .where(col(ApiKey.reserved_balance) >= deducted_max_cost)
                         .values(
-                            reserved_balance=col(ApiKey.reserved_balance) - deducted_max_cost,
-                            total_spent=col(ApiKey.total_spent) + min(billing_key.balance, total_cost_msats),
+                            reserved_balance=overrun_safe_reserved,
+                            total_spent=col(ApiKey.total_spent) + actual_charge_msats,
                         )
                     )
                     await session.exec(child_stmt)  # type: ignore[call-overload]
 
                 await session.commit()
 
-                if result.rowcount:
-                    await session.refresh(billing_key)
-                    if billing_key.hashed_key != key.hashed_key:
-                        await session.refresh(key)
-                    cost.total_msats = total_cost_msats
-                    logger.info(
-                        "Finalized payment with additional charge",
-                        extra={
-                            "key_hash": key.hashed_key[:8] + "...",
-                            "billing_key_hash": billing_key.hashed_key[:8] + "...",
-                            "charged_amount": total_cost_msats,
-                            "new_balance": billing_key.balance,
-                            "model": model,
-                        },
-                    )
-                    await _accumulate_fee(total_cost_msats)
-                    payments_logger.info(
-                        "FINALIZE",
-                        extra={
-                            "event": "finalize",
-                            "key_hash": key.hashed_key[:8] + "...",
-                            "billing_key_hash": billing_key.hashed_key[:8] + "...",
-                            "model": model,
-                            "cost_reserved": deducted_max_cost,
-                            "cost_charged": total_cost_msats,
-                            "input_tokens": cost.input_tokens,
-                            "output_tokens": cost.output_tokens,
-                            "balance": billing_key.balance,
-                            "reserved_balance": billing_key.reserved_balance,
-                            "total_spent": billing_key.total_spent,
-                            "finalize_type": "overrun",
-                        },
-                    )
-                else:
-                    # Guard fired: reservation was already released by a concurrent
-                    # finalization for this key. Nothing left to do.
-                    logger.warning(
-                        "Finalization skipped - reservation already released",
-                        extra={
-                            "key_hash": key.hashed_key[:8] + "...",
-                            "billing_key_hash": billing_key.hashed_key[:8] + "...",
-                            "attempted_charge": total_cost_msats,
-                            "model": model,
-                        },
-                    )
+                await session.refresh(billing_key)
+                if billing_key.hashed_key != key.hashed_key:
+                    await session.refresh(key)
+                cost.total_msats = actual_charge_msats
+                logger.info(
+                    "Finalized payment with additional charge",
+                    extra={
+                        "key_hash": key.hashed_key[:8] + "...",
+                        "billing_key_hash": billing_key.hashed_key[:8] + "...",
+                        "charged_amount": actual_charge_msats,
+                        "new_balance": billing_key.balance,
+                        "model": model,
+                    },
+                )
+                await _accumulate_fee(actual_charge_msats)
+                payments_logger.info(
+                    "FINALIZE",
+                    extra={
+                        "event": "finalize",
+                        "key_hash": key.hashed_key[:8] + "...",
+                        "billing_key_hash": billing_key.hashed_key[:8] + "...",
+                        "model": model,
+                        "cost_reserved": deducted_max_cost,
+                        "cost_charged": actual_charge_msats,
+                        "input_tokens": cost.input_tokens,
+                        "output_tokens": cost.output_tokens,
+                        "balance": billing_key.balance,
+                        "reserved_balance": billing_key.reserved_balance,
+                        "total_spent": billing_key.total_spent,
+                        "finalize_type": "overrun",
+                    },
+                )
             else:
                 # Refund some of the base cost
                 refund = abs(cost_difference)
@@ -1123,8 +1429,10 @@ async def adjust_payment_for_tokens(
                     )
 
                 refund_safe_reserved = case(
-                    (col(ApiKey.reserved_balance) >= deducted_max_cost,
-                     col(ApiKey.reserved_balance) - deducted_max_cost),
+                    (
+                        col(ApiKey.reserved_balance) >= deducted_max_cost,
+                        col(ApiKey.reserved_balance) - deducted_max_cost,
+                    ),
                     else_=0,
                 )
 
@@ -1142,8 +1450,10 @@ async def adjust_payment_for_tokens(
                 # Also update total_spent and reserved_balance on the child key if it's different
                 if billing_key.hashed_key != key.hashed_key:
                     child_refund_safe_reserved = case(
-                        (col(ApiKey.reserved_balance) >= deducted_max_cost,
-                         col(ApiKey.reserved_balance) - deducted_max_cost),
+                        (
+                            col(ApiKey.reserved_balance) >= deducted_max_cost,
+                            col(ApiKey.reserved_balance) - deducted_max_cost,
+                        ),
                         else_=0,
                     )
                     child_stmt = (
@@ -1298,6 +1608,33 @@ async def periodic_key_reset() -> None:
             logger.error(f"Error in periodic_key_reset: {e}")
 
 
+async def periodic_dead_key_prune() -> None:
+    """Periodically prune dead API keys. Interval <= 0 disables it.
+
+    See ``prune_dead_api_keys`` for eligibility.
+    """
+    from .core.db import create_session, prune_dead_api_keys
+
+    interval = settings.dead_key_prune_interval_seconds
+    if interval <= 0:
+        logger.info("Dead-key pruning disabled (interval <= 0)")
+        return
+
+    while True:
+        try:
+            await asyncio.sleep(interval)
+        except asyncio.CancelledError:
+            break
+
+        try:
+            async with create_session() as session:
+                await prune_dead_api_keys(session, settings.dead_key_min_age_seconds)
+        except asyncio.CancelledError:
+            break
+        except Exception as e:
+            logger.error(f"Error in periodic_dead_key_prune: {e}")
+
+
 STALE_RESERVATION_SWEEP_INTERVAL_SECONDS: int = 60
 
 

routstr/balance.py

@@ -7,7 +7,7 @@ from typing import Annotated, NoReturn
 from fastapi import APIRouter, Depends, Header, HTTPException
 from fastapi.responses import JSONResponse
 from pydantic import BaseModel
-from sqlmodel import col, or_, select, update
+from sqlmodel import col, select, update
 
 from .auth import get_billing_key, validate_bearer_key
 from .core.db import (
@@ -15,12 +15,22 @@ from .core.db import (
     AsyncSession,
     CashuTransaction,
     get_session,
-    store_cashu_transaction,
+    release_stale_reservations,
+)
+from .core.db import (
+    store_cashu_transaction_with_retry as store_cashu_transaction,
 )
 from .core.logging import get_logger
 from .core.settings import settings
 from .lightning import lightning_router
-from .wallet import credit_balance, recieve_token, send_to_lnurl, send_token
+from .wallet import (
+    classify_redemption_error,
+    credit_balance,
+    is_mint_connection_error,
+    recieve_token,
+    send_to_lnurl,
+    send_token,
+)
 
 router = APIRouter()
 balance_router = APIRouter(prefix="/v1/balance")
@@ -99,13 +109,19 @@ async def account_info(
 # Note: validate_bearer_key already supports refund_address and key_expiry_time params
 
 
-@router.get("/create")
-async def create_balance(
+class BalanceCreateRequest(BaseModel):
+    initial_balance_token: str
+    balance_limit: int | None = None
+    balance_limit_reset: str | None = None
+    validity_date: int | None = None
+
+
+async def _create_balance(
     initial_balance_token: str,
-    balance_limit: int | None = None,
-    balance_limit_reset: str | None = None,
-    validity_date: int | None = None,
-    session: AsyncSession = Depends(get_session),
+    balance_limit: int | None,
+    balance_limit_reset: str | None,
+    validity_date: int | None,
+    session: AsyncSession,
 ) -> dict:
     key = await validate_bearer_key(initial_balance_token, session)
 
@@ -125,6 +141,37 @@ async def create_balance(
     }
 
 
+@router.post("/create")
+async def create_balance_from_body(
+    payload: BalanceCreateRequest,
+    session: AsyncSession = Depends(get_session),
+) -> dict:
+    return await _create_balance(
+        payload.initial_balance_token,
+        payload.balance_limit,
+        payload.balance_limit_reset,
+        payload.validity_date,
+        session,
+    )
+
+
+@router.get("/create")
+async def create_balance(
+    initial_balance_token: str,
+    balance_limit: int | None = None,
+    balance_limit_reset: str | None = None,
+    validity_date: int | None = None,
+    session: AsyncSession = Depends(get_session),
+) -> dict:
+    return await _create_balance(
+        initial_balance_token,
+        balance_limit,
+        balance_limit_reset,
+        validity_date,
+        session,
+    )
+
+
 @router.get("/info")
 async def wallet_info(
     key: ApiKey = Depends(get_key_from_header),
@@ -156,30 +203,18 @@ async def topup_wallet_endpoint(
         raise HTTPException(status_code=400, detail="Invalid token format")
     try:
         amount_msats = await credit_balance(cashu_token, billing_key, session)
-    except ValueError as e:
-        error_msg = str(e)
-        if "already spent" in error_msg.lower():
-            raise HTTPException(status_code=400, detail="Token already spent")
-        elif "invalid" in error_msg.lower() or "decode" in error_msg.lower():
-            raise HTTPException(status_code=400, detail="Invalid token format")
-        elif "insufficient" in error_msg.lower() or "melt fee" in error_msg.lower():
-            raise HTTPException(
-                status_code=400,
-                detail=f"Token value is too small to cover swap fees. {error_msg}",
-            )
-        elif "failed to melt" in error_msg.lower():
-            raise HTTPException(
-                status_code=400,
-                detail=f"Failed to swap foreign mint token. {error_msg}",
-            )
-        else:
-            raise HTTPException(status_code=400, detail=f"Failed to redeem token: {error_msg}")
     except Exception as e:
-        logger.error(
-            "topup_wallet_endpoint: unhandled error",
-            extra={"error": str(e), "error_type": type(e).__name__},
-        )
-        raise HTTPException(status_code=500, detail="Internal server error")
+        # Shared taxonomy so top-up matches the bearer/X-Cashu paths (503 for an
+        # unreachable mint, 422 for fee/swap failures, 400 for token faults).
+        classified = classify_redemption_error(e)
+        if classified is None:
+            logger.error(
+                "topup_wallet_endpoint: unhandled error",
+                extra={"error": str(e), "error_type": type(e).__name__},
+            )
+            raise HTTPException(status_code=500, detail="Internal server error")
+        _type, status_code, message, _code = classified
+        raise HTTPException(status_code=status_code, detail=message)
     return {"msats": amount_msats}
 
 
@@ -274,7 +309,20 @@ async def refund_wallet_endpoint(
         )
         out_tx = out_tx_result.first()
         if out_tx is None:
-            raise HTTPException(status_code=404, detail="Refund not found")
+            # The "in" row exists with a request_id, but the "out" (refund)
+            # row hasn't been written yet — the upstream request is still in
+            # flight and the refund will be minted once it completes. Tell the
+            # client to retry instead of 404ing permanently (race condition
+            # where /v1/wallet/refund is polled before the refund exists).
+            logger.debug(
+                "refund_wallet_endpoint: refund pending (in row exists, out row not yet created)",
+                extra={"request_id": in_tx.request_id},
+            )
+            raise HTTPException(
+                status_code=425,
+                detail="Refund is pending; retry shortly.",
+                headers={"Retry-After": "2"},
+            )
         if out_tx.swept:
             raise HTTPException(status_code=410, detail="Refund has been swept")
 
@@ -313,30 +361,19 @@ async def refund_wallet_endpoint(
         )
 
     if key.reserved_balance > 0:
-        # Release the reservation if it is stale
-        cutoff = int(time.time()) - settings.stale_reservation_timeout_seconds
-        stale_release_stmt = (
-            update(ApiKey)
-            .where(col(ApiKey.hashed_key) == key.hashed_key)
-            .where(col(ApiKey.reserved_balance) > 0)
-            .where(
-                or_(
-                    col(ApiKey.reserved_at).is_(None),
-                    col(ApiKey.reserved_at) < cutoff,
-                )
-            )
-            .values(reserved_balance=0, reserved_at=None)
+        # Release only durable reservations old enough to be stale. A newer
+        # request on the same aggregate balance must remain reserved.
+        await release_stale_reservations(
+            session,
+            settings.stale_reservation_timeout_seconds,
+            key_hash=key.hashed_key,
         )
-        stale_result = await session.exec(stale_release_stmt)  # type: ignore[call-overload]
-        await session.commit()
-
-        if stale_result.rowcount == 0:
+        await session.refresh(key)
+        if key.reserved_balance > 0:
             raise HTTPException(
                 status_code=400,
                 detail="Cannot refund key. There are ongoing requests for this api key.",
             )
-
-        await session.refresh(key)
         logger.warning(
             "refund_wallet_endpoint: released stale reservation before refund",
             extra={
@@ -441,14 +478,10 @@ async def refund_wallet_endpoint(
                 "has_refund_address": bool(key.refund_address),
             },
         )
-        if (
-            "mint" in error_msg.lower()
-            or "connection" in error_msg.lower()
-            or "ConnectError" in str(type(e))
-        ):
-            raise HTTPException(status_code=503, detail=f"Mint service unavailable: {error_msg}")
+        if is_mint_connection_error(e):
+            raise HTTPException(status_code=503, detail="Mint service unavailable")
         else:
-            raise HTTPException(status_code=500, detail=f"Refund failed: {error_msg}")
+            raise HTTPException(status_code=500, detail="Refund failed")
 
     await _refund_cache_set(bearer_value, result)
 

routstr/core/admin.py

@@ -1,5 +1,6 @@
 import asyncio
 import json
+import re
 import secrets
 from datetime import datetime, timezone
 from pathlib import Path
@@ -8,6 +9,7 @@ from fastapi import APIRouter, Depends, HTTPException, Query, Request
 from pydantic import BaseModel, RootModel
 from pydantic.v1 import ValidationError as PydanticValidationError
 from sqlmodel import select
+from sqlmodel.ext.asyncio.session import AsyncSession
 
 from ..payment.models import _row_to_model, list_models
 from ..proxy import refresh_model_maps, reinitialize_upstreams
@@ -18,6 +20,7 @@ from ..wallet import (
     send_token,
     slow_filter_spend_proofs,
 )
+from . import vault
 from .db import (
     ApiKey,
     CashuTransaction,
@@ -26,10 +29,17 @@ from .db import (
     ModelRow,
     UpstreamProviderRow,
     create_session,
+    get_secret,
+    set_admin_password,
+    set_nsec,
+)
+from .db import (
+    store_cashu_transaction_with_retry as store_cashu_transaction,
 )
 from .log_manager import log_manager
 from .logging import get_logger
-from .settings import SettingsService, settings
+from .provider_slugs import allocate_unique_provider_slug
+from .settings import SettingsService, derive_npub_from_nsec, settings
 
 logger = get_logger(__name__)
 
@@ -200,8 +210,6 @@ async def get_settings(request: Request) -> dict:
     data = settings.dict()
     if "upstream_api_key" in data:
         data["upstream_api_key"] = "[REDACTED]" if data["upstream_api_key"] else ""
-    if "admin_password" in data:
-        data["admin_password"] = "[REDACTED]" if data["admin_password"] else ""
     if "nsec" in data:
         data["nsec"] = "[REDACTED]" if data["nsec"] else ""
     return data
@@ -218,9 +226,10 @@ class PasswordUpdate(BaseModel):
 
 @admin_router.patch("/api/settings", dependencies=[Depends(require_admin_api)])
 async def update_settings(request: Request, update: SettingsUpdate) -> dict:
-    # Remove sensitive fields from general settings update
+    # Secrets are not editable through the general settings endpoint; they have
+    # dedicated rotation paths and never reach the settings blob.
     settings_data = update.root.copy()
-    sensitive_fields = ["admin_password", "upstream_api_key", "nsec"]
+    sensitive_fields = ["upstream_api_key", "nsec"]
     for field in sensitive_fields:
         if field in settings_data:
             del settings_data[field]
@@ -235,8 +244,6 @@ async def update_settings(request: Request, update: SettingsUpdate) -> dict:
     data = new_settings.dict()
     if "upstream_api_key" in data:
         data["upstream_api_key"] = "[REDACTED]" if data["upstream_api_key"] else ""
-    if "admin_password" in data:
-        data["admin_password"] = "[REDACTED]" if data["admin_password"] else ""
     if "nsec" in data:
         data["nsec"] = "[REDACTED]" if data["nsec"] else ""
     return data
@@ -244,44 +251,63 @@ async def update_settings(request: Request, update: SettingsUpdate) -> dict:
 
 @admin_router.patch("/api/password", dependencies=[Depends(require_admin_api)])
 async def update_password(request: Request, password_update: PasswordUpdate) -> dict:
-    current_password = settings.admin_password
-
-    if not current_password:
-        raise HTTPException(status_code=500, detail="Admin password not configured")
-
-    if password_update.current_password != current_password:
-        raise HTTPException(status_code=401, detail="Current password is incorrect")
-
-    # Validate new password
-    new_password = password_update.new_password.strip()
-    if len(new_password) < 6:
-        raise HTTPException(
-            status_code=400, detail="New password must be at least 6 characters"
-        )
-
-    # Update password
     async with create_session() as session:
-        await SettingsService.update({"admin_password": new_password}, session)
+        secret = await get_secret(session)
+
+        if not secret.admin_password_hash:
+            raise HTTPException(
+                status_code=500, detail="Admin password not configured"
+            )
+
+        if not vault.verify_password(
+            password_update.current_password, secret.admin_password_hash
+        ):
+            raise HTTPException(
+                status_code=401, detail="Current password is incorrect"
+            )
+
+        # Validate new password
+        new_password = password_update.new_password.strip()
+        if len(new_password) < vault.MIN_PASSWORD_LENGTH:
+            raise HTTPException(
+                status_code=400,
+                detail=(
+                    "New password must be at least "
+                    f"{vault.MIN_PASSWORD_LENGTH} characters"
+                ),
+            )
+
+        await set_admin_password(session, new_password)
 
     return {"ok": True, "message": "Password updated successfully"}
 
 
-class SetupRequest(BaseModel):
-    password: str
+class NsecUpdate(BaseModel):
+    nsec: str
 
 
-@admin_router.post("/api/setup")
-async def initial_setup(request: Request, payload: SetupRequest) -> dict[str, object]:
-    if settings.admin_password:
-        raise HTTPException(status_code=409, detail="Admin password already set")
-    pw = (payload.password or "").strip()
-    if len(pw) < 8:
-        raise HTTPException(
-            status_code=400, detail="Password must be at least 8 characters"
-        )
+@admin_router.patch("/api/nsec", dependencies=[Depends(require_admin_api)])
+async def update_nsec(request: Request, payload: NsecUpdate) -> dict[str, object]:
+    # The node's Nostr identity is a secret: it is stored encrypted in the
+    # Secret store, never in the settings blob, so it gets its own endpoint
+    # rather than riding the general settings PATCH (which strips it). An empty
+    # nsec clears the identity.
+    nsec = payload.nsec.strip()
+    npub = ""
+    if nsec:
+        derived = derive_npub_from_nsec(nsec)
+        if not derived:
+            raise HTTPException(status_code=400, detail="Invalid nsec")
+        npub = derived
+
     async with create_session() as session:
-        await SettingsService.update({"admin_password": pw}, session)
-    return {"ok": True}
+        await set_nsec(session, nsec)
+
+    # Reflect the change in the live runtime so Nostr signing/announcements pick
+    # it up without a restart (mirrors what bootstrap_secrets sets at boot).
+    settings.nsec = nsec
+    settings.npub = npub
+    return {"ok": True, "npub": npub}
 
 
 class AdminLoginRequest(BaseModel):
@@ -292,12 +318,16 @@ class AdminLoginRequest(BaseModel):
 async def admin_login(
     request: Request, payload: AdminLoginRequest
 ) -> dict[str, object]:
-    admin_pw = settings.admin_password
+    async with create_session() as session:
+        secret = await get_secret(session)
+        # Read the hash while the session is open; the ORM object is detached
+        # once the context exits and its attributes can no longer be loaded.
+        password_hash = secret.admin_password_hash
 
-    if not admin_pw:
+    if not password_hash:
         raise HTTPException(status_code=500, detail="Admin password not configured")
 
-    if payload.password != admin_pw:
+    if not vault.verify_password(payload.password, password_hash):
         raise HTTPException(status_code=401, detail="Invalid password")
 
     token = secrets.token_urlsafe(32)
@@ -408,12 +438,11 @@ async def withdraw(
     # Get wallet and check balance
     from .settings import settings as global_settings
 
-    wallet = await get_wallet(
-        withdraw_request.mint_url or global_settings.primary_mint, withdraw_request.unit
-    )
+    effective_mint = withdraw_request.mint_url or global_settings.primary_mint
+    wallet = await get_wallet(effective_mint, withdraw_request.unit)
     proofs = get_proofs_per_mint_and_unit(
         wallet,
-        withdraw_request.mint_url or global_settings.primary_mint,
+        effective_mint,
         withdraw_request.unit,
         not_reserved=True,
     )
@@ -429,8 +458,27 @@ async def withdraw(
         raise HTTPException(status_code=400, detail="Insufficient wallet balance")
 
     token = await send_token(
-        withdraw_request.amount, withdraw_request.unit, withdraw_request.mint_url
+        withdraw_request.amount, withdraw_request.unit, effective_mint
     )
+    try:
+        await store_cashu_transaction(
+            token=token,
+            amount=withdraw_request.amount,
+            unit=withdraw_request.unit,
+            mint_url=effective_mint,
+            typ="out",
+            collected=False,
+            source="admin",
+        )
+    except Exception:
+        logger.critical(
+            "Admin withdrawal token issued without a persisted audit record",
+            extra={
+                "amount": withdraw_request.amount,
+                "unit": withdraw_request.unit,
+                "mint_url": effective_mint,
+            },
+        )
     return {"token": token}
 
 
@@ -456,19 +504,17 @@ class ModelCreate(BaseModel):
     dependencies=[Depends(require_admin_api)],
 )
 async def upsert_provider_model(
-    provider_id: int, payload: ModelCreate
+    provider_id: str, payload: ModelCreate
 ) -> dict[str, object]:
-    print(payload)
     logger.info(
         f"UPSERT_PROVIDER_MODEL called: provider_id={provider_id}, model_id={payload.id}"
     )
     async with create_session() as session:
-        provider = await session.get(UpstreamProviderRow, provider_id)
-        if not provider:
-            raise HTTPException(status_code=404, detail="Provider not found")
+        provider = await _get_upstream_provider_by_ref(session, provider_id)
+        provider_pk = _provider_pk(provider)
 
         # Try to get existing model
-        existing_row = await session.get(ModelRow, (payload.id, provider_id))
+        existing_row = await session.get(ModelRow, (payload.id, provider_pk))
 
         if existing_row:
             # Update existing model
@@ -524,7 +570,7 @@ async def upsert_provider_model(
                 alias_ids=(
                     json.dumps(payload.alias_ids) if payload.alias_ids else None
                 ),
-                upstream_provider_id=provider_id,
+                upstream_provider_id=provider_pk,
                 enabled=payload.enabled,
                 forwarded_model_id=payload.forwarded_model_id or payload.id,
             )
@@ -543,7 +589,7 @@ async def upsert_provider_model(
     dependencies=[Depends(require_admin_api)],
 )
 async def update_provider_model_legacy(
-    provider_id: int, model_id: str, payload: ModelCreate
+    provider_id: str, model_id: str, payload: ModelCreate
 ) -> dict[str, object]:
     """Legacy PATCH endpoint - redirects to upsert POST endpoint for backward compatibility."""
     logger.info(
@@ -556,13 +602,12 @@ async def update_provider_model_legacy(
     "/api/upstream-providers/{provider_id}/models/{model_id:path}",
     dependencies=[Depends(require_admin_api)],
 )
-async def get_provider_model(provider_id: int, model_id: str) -> dict[str, object]:
+async def get_provider_model(provider_id: str, model_id: str) -> dict[str, object]:
     async with create_session() as session:
-        provider = await session.get(UpstreamProviderRow, provider_id)
-        if not provider:
-            raise HTTPException(status_code=404, detail="Provider not found")
+        provider = await _get_upstream_provider_by_ref(session, provider_id)
+        provider_pk = _provider_pk(provider)
 
-        row = await session.get(ModelRow, (model_id, provider_id))
+        row = await session.get(ModelRow, (model_id, provider_pk))
         if not row:
             raise HTTPException(
                 status_code=404, detail="Model not found for this provider"
@@ -576,9 +621,11 @@ async def get_provider_model(provider_id: int, model_id: str) -> dict[str, objec
     "/api/upstream-providers/{provider_id}/models/{model_id:path}",
     dependencies=[Depends(require_admin_api)],
 )
-async def delete_provider_model(provider_id: int, model_id: str) -> dict[str, object]:
+async def delete_provider_model(provider_id: str, model_id: str) -> dict[str, object]:
     async with create_session() as session:
-        row = await session.get(ModelRow, (model_id, provider_id))
+        provider = await _get_upstream_provider_by_ref(session, provider_id)
+        provider_pk = _provider_pk(provider)
+        row = await session.get(ModelRow, (model_id, provider_pk))
         if not row:
             raise HTTPException(
                 status_code=404, detail="Model not found for this provider"
@@ -593,10 +640,12 @@ async def delete_provider_model(provider_id: int, model_id: str) -> dict[str, ob
     "/api/upstream-providers/{provider_id}/models",
     dependencies=[Depends(require_admin_api)],
 )
-async def delete_all_provider_models(provider_id: int) -> dict[str, object]:
+async def delete_all_provider_models(provider_id: str) -> dict[str, object]:
     async with create_session() as session:
+        provider = await _get_upstream_provider_by_ref(session, provider_id)
+        provider_pk = _provider_pk(provider)
         result = await session.exec(
-            select(ModelRow).where(ModelRow.upstream_provider_id == provider_id)
+            select(ModelRow).where(ModelRow.upstream_provider_id == provider_pk)
         )  # type: ignore
         rows = result.all()
         for row in rows:
@@ -615,7 +664,7 @@ class BatchOverrideRequest(BaseModel):
     dependencies=[Depends(require_admin_api)],
 )
 async def batch_override_provider_models(
-    provider_id: int, payload: BatchOverrideRequest
+    provider_id: str, payload: BatchOverrideRequest
 ) -> dict[str, object]:
     """Batch override models for a specific provider."""
     logger.info(
@@ -623,15 +672,14 @@ async def batch_override_provider_models(
     )
 
     async with create_session() as session:
-        provider = await session.get(UpstreamProviderRow, provider_id)
-        if not provider:
-            raise HTTPException(status_code=404, detail="Provider not found")
+        provider = await _get_upstream_provider_by_ref(session, provider_id)
+        provider_pk = _provider_pk(provider)
 
         overridden_count = 0
 
         for model_data in payload.models:
             # Try to get existing model regardless of whether it's enabled or not
-            existing_row = await session.get(ModelRow, (model_data.id, provider_id))
+            existing_row = await session.get(ModelRow, (model_data.id, provider_pk))
 
             if existing_row:
                 # Update existing
@@ -685,7 +733,7 @@ async def batch_override_provider_models(
                         if model_data.alias_ids
                         else None
                     ),
-                    upstream_provider_id=provider_id,
+                    upstream_provider_id=provider_pk,
                     enabled=model_data.enabled,
                 )
                 session.add(row)
@@ -702,6 +750,85 @@ async def batch_override_provider_models(
     }
 
 
+_SLUG_PATTERN = re.compile(r"^[a-z0-9][a-z0-9-]{1,62}[a-z0-9]$")
+
+
+def _validate_slug(value: str) -> str:
+    candidate = value.strip().lower()
+    if not _SLUG_PATTERN.fullmatch(candidate):
+        raise HTTPException(
+            status_code=400,
+            detail=(
+                "slug must be 3-64 chars, lowercase letters/digits/hyphens, "
+                "and may not start or end with a hyphen"
+            ),
+        )
+    if candidate.isdigit():
+        raise HTTPException(
+            status_code=400,
+            detail="slug must not be all digits",
+        )
+    return candidate
+
+
+async def _ensure_unique_slug(
+    session: AsyncSession, slug: str, exclude_id: int | None = None
+) -> None:
+    stmt = select(UpstreamProviderRow).where(UpstreamProviderRow.slug == slug)
+    result = await session.exec(stmt)
+    existing = result.first()
+    if existing and existing.id != exclude_id:
+        raise HTTPException(
+            status_code=409,
+            detail="Provider with this slug already exists",
+        )
+
+
+async def _get_upstream_provider_by_ref(
+    session: AsyncSession, provider_ref: str
+) -> UpstreamProviderRow:
+    if provider_ref.isdigit():
+        provider = await session.get(UpstreamProviderRow, int(provider_ref))
+    else:
+        slug = _validate_slug(provider_ref)
+        result = await session.exec(
+            select(UpstreamProviderRow).where(UpstreamProviderRow.slug == slug)
+        )
+        provider = result.first()
+
+    if not provider:
+        raise HTTPException(status_code=404, detail="Provider not found")
+    return provider
+
+
+def _provider_pk(provider: UpstreamProviderRow) -> int:
+    if provider.id is None:
+        raise HTTPException(status_code=500, detail="Provider has no database id")
+    return provider.id
+
+
+def _serialize_provider(
+    provider: UpstreamProviderRow, redact_api_key: bool = True
+) -> dict[str, object]:
+    return {
+        "id": provider.id,
+        "slug": provider.slug,
+        "provider_type": provider.provider_type,
+        "base_url": provider.base_url,
+        "api_key": "[REDACTED]"
+        if (redact_api_key and provider.api_key)
+        else provider.api_key
+        if not redact_api_key
+        else "",
+        "api_version": provider.api_version,
+        "enabled": provider.enabled,
+        "provider_fee": provider.provider_fee,
+        "provider_settings": json.loads(provider.provider_settings)
+        if provider.provider_settings
+        else None,
+    }
+
+
 class UpstreamProviderCreate(BaseModel):
     provider_type: str
     base_url: str
@@ -710,6 +837,7 @@ class UpstreamProviderCreate(BaseModel):
     enabled: bool = True
     provider_fee: float = 1.01
     provider_settings: dict | None = None
+    slug: str | None = None
 
 
 class UpstreamProviderUpdate(BaseModel):
@@ -720,6 +848,50 @@ class UpstreamProviderUpdate(BaseModel):
     enabled: bool | None = None
     provider_fee: float | None = None
     provider_settings: dict | None = None
+    slug: str | None = None
+
+
+class UpstreamProviderUpdateBySlug(BaseModel):
+    slug: str
+    new_slug: str | None = None
+    provider_type: str | None = None
+    base_url: str | None = None
+    api_key: str | None = None
+    api_version: str | None = None
+    enabled: bool | None = None
+    provider_fee: float | None = None
+    provider_settings: dict | None = None
+
+
+async def _apply_provider_update(
+    session: AsyncSession,
+    provider: UpstreamProviderRow,
+    payload: UpstreamProviderUpdate,
+    new_slug: str | None = None,
+) -> None:
+    if new_slug is not None:
+        validated = _validate_slug(new_slug)
+        await _ensure_unique_slug(session, validated, exclude_id=provider.id)
+        provider.slug = validated
+
+    if payload.provider_type is not None:
+        provider.provider_type = payload.provider_type
+    if payload.base_url is not None:
+        provider.base_url = payload.base_url
+    if payload.api_key is not None:
+        provider.api_key = payload.api_key
+    if payload.api_version is not None:
+        provider.api_version = payload.api_version
+    if payload.enabled is not None:
+        provider.enabled = payload.enabled
+    if payload.provider_fee is not None:
+        provider.provider_fee = payload.provider_fee
+    if payload.provider_settings is not None:
+        provider.provider_settings = json.dumps(payload.provider_settings)
+
+    session.add(provider)
+    await session.commit()
+    await session.refresh(provider)
 
 
 @admin_router.get("/api/upstream-providers", dependencies=[Depends(require_admin_api)])
@@ -727,21 +899,7 @@ async def get_upstream_providers() -> list[dict[str, object]]:
     async with create_session() as session:
         result = await session.exec(select(UpstreamProviderRow))
         providers = result.all()
-        return [
-            {
-                "id": p.id,
-                "provider_type": p.provider_type,
-                "base_url": p.base_url,
-                "api_key": "[REDACTED]" if p.api_key else "",
-                "api_version": p.api_version,
-                "enabled": p.enabled,
-                "provider_fee": p.provider_fee,
-                "provider_settings": json.loads(p.provider_settings)
-                if p.provider_settings
-                else None,
-            }
-            for p in providers
-        ]
+        return [_serialize_provider(p) for p in providers]
 
 
 @admin_router.post("/api/upstream-providers", dependencies=[Depends(require_admin_api)])
@@ -761,7 +919,14 @@ async def create_upstream_provider(
                 detail="Provider with this base URL and API key already exists",
             )
 
+        if payload.slug:
+            slug = _validate_slug(payload.slug)
+            await _ensure_unique_slug(session, slug)
+        else:
+            slug = await allocate_unique_provider_slug(session, payload.provider_type)
+
         provider = UpstreamProviderRow(
+            slug=slug,
             provider_type=payload.provider_type,
             base_url=payload.base_url,
             api_key=payload.api_key,
@@ -778,99 +943,81 @@ async def create_upstream_provider(
 
     await reinitialize_upstreams()
     await refresh_model_maps()
-    return {
-        "id": provider.id,
-        "provider_type": provider.provider_type,
-        "base_url": provider.base_url,
-        "api_key": "[REDACTED]",
-        "api_version": provider.api_version,
-        "enabled": provider.enabled,
-        "provider_fee": provider.provider_fee,
-        "provider_settings": payload.provider_settings,
-    }
+    return _serialize_provider(provider)
 
 
 @admin_router.get(
     "/api/upstream-providers/{provider_id}", dependencies=[Depends(require_admin_api)]
 )
-async def get_upstream_provider(provider_id: int) -> dict[str, object]:
+async def get_upstream_provider(provider_id: str) -> dict[str, object]:
     async with create_session() as session:
-        provider = await session.get(UpstreamProviderRow, provider_id)
-        if not provider:
-            raise HTTPException(status_code=404, detail="Provider not found")
-        return {
-            "id": provider.id,
-            "provider_type": provider.provider_type,
-            "base_url": provider.base_url,
-            "api_key": "[REDACTED]" if provider.api_key else "",
-            "api_version": provider.api_version,
-            "enabled": provider.enabled,
-            "provider_fee": provider.provider_fee,
-            "provider_settings": json.loads(provider.provider_settings)
-            if provider.provider_settings
-            else None,
-        }
+        provider = await _get_upstream_provider_by_ref(session, provider_id)
+        return _serialize_provider(provider)
 
 
 @admin_router.patch(
     "/api/upstream-providers/{provider_id}", dependencies=[Depends(require_admin_api)]
 )
 async def update_upstream_provider(
-    provider_id: int, payload: UpstreamProviderUpdate
+    provider_id: str, payload: UpstreamProviderUpdate
 ) -> dict[str, object]:
     async with create_session() as session:
-        provider = await session.get(UpstreamProviderRow, provider_id)
-        if not provider:
-            raise HTTPException(status_code=404, detail="Provider not found")
+        provider = await _get_upstream_provider_by_ref(session, provider_id)
 
-        if payload.provider_type is not None:
-            provider.provider_type = payload.provider_type
-        if payload.base_url is not None:
-            provider.base_url = payload.base_url
-        if payload.api_key is not None:
-            provider.api_key = payload.api_key
-        if payload.api_version is not None:
-            provider.api_version = payload.api_version
-        if payload.enabled is not None:
-            provider.enabled = payload.enabled
-        if payload.provider_fee is not None:
-            provider.provider_fee = payload.provider_fee
-        if payload.provider_settings is not None:
-            provider.provider_settings = json.dumps(payload.provider_settings)
-
-        session.add(provider)
-        await session.commit()
-        await session.refresh(provider)
+        await _apply_provider_update(session, provider, payload, new_slug=payload.slug)
 
     await reinitialize_upstreams()
     await refresh_model_maps()
-    return {
-        "id": provider.id,
-        "provider_type": provider.provider_type,
-        "base_url": provider.base_url,
-        "api_key": "[REDACTED]",
-        "api_version": provider.api_version,
-        "enabled": provider.enabled,
-        "provider_fee": provider.provider_fee,
-        "provider_settings": json.loads(provider.provider_settings)
-        if provider.provider_settings
-        else None,
-    }
+    return _serialize_provider(provider)
+
+
+@admin_router.patch(
+    "/api/upstream-providers", dependencies=[Depends(require_admin_api)]
+)
+async def update_upstream_provider_by_slug(
+    payload: UpstreamProviderUpdateBySlug,
+) -> dict[str, object]:
+    lookup = _validate_slug(payload.slug)
+    async with create_session() as session:
+        result = await session.exec(
+            select(UpstreamProviderRow).where(
+                UpstreamProviderRow.slug == lookup
+            )
+        )
+        provider = result.first()
+        if not provider:
+            raise HTTPException(status_code=404, detail="Provider not found")
+
+        update_payload = UpstreamProviderUpdate(
+            provider_type=payload.provider_type,
+            base_url=payload.base_url,
+            api_key=payload.api_key,
+            api_version=payload.api_version,
+            enabled=payload.enabled,
+            provider_fee=payload.provider_fee,
+            provider_settings=payload.provider_settings,
+        )
+        await _apply_provider_update(
+            session, provider, update_payload, new_slug=payload.new_slug
+        )
+
+    await reinitialize_upstreams()
+    await refresh_model_maps()
+    return _serialize_provider(provider)
 
 
 @admin_router.delete(
     "/api/upstream-providers/{provider_id}", dependencies=[Depends(require_admin_api)]
 )
-async def delete_upstream_provider(provider_id: int) -> dict[str, object]:
+async def delete_upstream_provider(provider_id: str) -> dict[str, object]:
     async with create_session() as session:
-        provider = await session.get(UpstreamProviderRow, provider_id)
-        if not provider:
-            raise HTTPException(status_code=404, detail="Provider not found")
+        provider = await _get_upstream_provider_by_ref(session, provider_id)
+        deleted_id = _provider_pk(provider)
         await session.delete(provider)
         await session.commit()
     await reinitialize_upstreams()
     await refresh_model_maps()
-    return {"ok": True, "deleted_id": provider_id}
+    return {"ok": True, "deleted_id": deleted_id}
 
 
 @admin_router.get("/api/provider-types", dependencies=[Depends(require_admin_api)])
@@ -885,17 +1032,16 @@ async def get_provider_types() -> list[dict[str, object]]:
     "/api/upstream-providers/{provider_id}/models",
     dependencies=[Depends(require_admin_api)],
 )
-async def get_provider_models(provider_id: int) -> dict[str, object]:
+async def get_provider_models(provider_id: str) -> dict[str, object]:
     from ..upstream.helpers import _instantiate_provider
 
     async with create_session() as session:
-        provider = await session.get(UpstreamProviderRow, provider_id)
-        if not provider:
-            raise HTTPException(status_code=404, detail="Provider not found")
+        provider = await _get_upstream_provider_by_ref(session, provider_id)
+        provider_pk = _provider_pk(provider)
 
         db_models = await list_models(
             session=session,
-            upstream_id=provider_id,
+            upstream_id=provider_pk,
             include_disabled=True,
             apply_fees=False,
         )
@@ -985,13 +1131,11 @@ class TopupTokenRequest(BaseModel):
     dependencies=[Depends(require_admin_api)],
 )
 async def topup_provider_with_token(
-    provider_id: int, payload: TopupTokenRequest
+    provider_id: str, payload: TopupTokenRequest
 ) -> dict:
     """Redeem a Cashu token for an upstream provider."""
     async with create_session() as session:
-        provider = await session.get(UpstreamProviderRow, provider_id)
-        if not provider:
-            raise HTTPException(status_code=404, detail="Provider not found")
+        provider = await _get_upstream_provider_by_ref(session, provider_id)
 
         import httpx
 
@@ -1022,15 +1166,13 @@ async def topup_provider_with_token(
     dependencies=[Depends(require_admin_api)],
 )
 async def initiate_provider_topup(
-    provider_id: int, payload: TopupRequest
+    provider_id: str, payload: TopupRequest
 ) -> dict[str, object]:
     """Initiate a Lightning Network top-up for the upstream provider account."""
     from ..upstream.helpers import _instantiate_provider
 
     async with create_session() as session:
-        provider = await session.get(UpstreamProviderRow, provider_id)
-        if not provider:
-            raise HTTPException(status_code=404, detail="Provider not found")
+        provider = await _get_upstream_provider_by_ref(session, provider_id)
 
         try:
             logger.info(
@@ -1150,15 +1292,13 @@ async def initiate_provider_topup(
     "/api/upstream-providers/{provider_id}/topup/{invoice_id}/status",
     dependencies=[Depends(require_admin_api)],
 )
-async def check_topup_status(provider_id: int, invoice_id: str) -> dict[str, object]:
+async def check_topup_status(provider_id: str, invoice_id: str) -> dict[str, object]:
     """Check the status of a Lightning Network top-up invoice."""
     from ..upstream.helpers import _instantiate_provider
     from ..upstream.ppqai import PPQAIUpstreamProvider
 
     async with create_session() as session:
-        provider = await session.get(UpstreamProviderRow, provider_id)
-        if not provider:
-            raise HTTPException(status_code=404, detail="Provider not found")
+        provider = await _get_upstream_provider_by_ref(session, provider_id)
 
         # For Routstr providers, proxy the status check
         if provider.provider_type == "routstr":
@@ -1205,14 +1345,12 @@ async def check_topup_status(provider_id: int, invoice_id: str) -> dict[str, obj
     "/api/upstream-providers/{provider_id}/balance",
     dependencies=[Depends(require_admin_api)],
 )
-async def get_provider_balance(provider_id: int) -> dict[str, object]:
+async def get_provider_balance(provider_id: str) -> dict[str, object]:
     """Get the current balance for an upstream provider account."""
     from ..upstream.helpers import _instantiate_provider
 
     async with create_session() as session:
-        provider = await session.get(UpstreamProviderRow, provider_id)
-        if not provider:
-            raise HTTPException(status_code=404, detail="Provider not found")
+        provider = await _get_upstream_provider_by_ref(session, provider_id)
 
         # For Routstr providers, proxy the balance check
         if provider.provider_type == "routstr":
@@ -1531,12 +1669,18 @@ async def get_transactions_api(
         )
         total = count_result.one()
 
-        stmt = base.order_by(col(CashuTransaction.created_at).desc()).offset(offset).limit(limit)
+        stmt = (
+            base.order_by(col(CashuTransaction.created_at).desc())
+            .offset(offset)
+            .limit(limit)
+        )
         results = await session.exec(stmt)
         transactions = results.all()
 
         return {
-            "transactions": [tx.dict() for tx in transactions],
+            "transactions": [
+                tx.dict(exclude={"sweep_started_at"}) for tx in transactions
+            ],
             "total": total,
         }
 
@@ -1591,15 +1735,13 @@ async def get_lightning_invoices_api(
     "/api/upstream-providers/{provider_id}/routstr/refund",
     dependencies=[Depends(require_admin_api)],
 )
-async def refund_routstr_provider_balance(provider_id: int) -> dict[str, object]:
+async def refund_routstr_provider_balance(provider_id: str) -> dict[str, object]:
     """Refund balance from an upstream Routstr provider back to the local wallet."""
     from ..upstream.helpers import _instantiate_provider
     from ..upstream.routstr import RoutstrUpstreamProvider
 
     async with create_session() as session:
-        provider_row = await session.get(UpstreamProviderRow, provider_id)
-        if not provider_row:
-            raise HTTPException(status_code=404, detail="Provider not found")
+        provider_row = await _get_upstream_provider_by_ref(session, provider_id)
 
         if provider_row.provider_type != "routstr":
             raise HTTPException(

routstr/core/db.py

@@ -1,28 +1,91 @@
+import asyncio
+import hashlib
 import os
 import pathlib
 import sqlite3
 import time
 import uuid
 from contextlib import asynccontextmanager
+from enum import Enum
 from typing import AsyncGenerator
 
 from alembic import command
 from alembic.config import Config
 from alembic.util.exc import CommandError
-from sqlalchemy import UniqueConstraint
-from sqlalchemy.exc import OperationalError
+from sqlalchemy import Index, UniqueConstraint, case, delete, event, or_
+from sqlalchemy.engine import make_url
+from sqlalchemy.exc import IntegrityError, OperationalError
+from sqlalchemy.ext.asyncio import AsyncEngine
 from sqlalchemy.ext.asyncio.engine import create_async_engine
+from sqlalchemy.orm import aliased
 from sqlmodel import Field, Relationship, SQLModel, col, func, select, update
 from sqlmodel.ext.asyncio.session import AsyncSession
 
 from .logging import get_logger
+from .settings import settings
 
 logger = get_logger(__name__)
 
 DATABASE_URL = os.environ.get("DATABASE_URL", "sqlite+aiosqlite:///keys.db")
 
 
-engine = create_async_engine(DATABASE_URL, echo=False)  # echo=True for debugging SQL
+def create_db_engine(database_url: str = DATABASE_URL) -> AsyncEngine:
+    """Build and instrument an async engine from environment-only settings."""
+    url = make_url(database_url)
+    backend = url.get_backend_name()
+    is_sqlite = backend == "sqlite"
+    is_memory_sqlite = is_sqlite and url.database in {None, "", ":memory:"}
+    pool_pre_ping = settings.database_pool_pre_ping or not is_sqlite
+    options: dict[str, int | float | bool] = {"pool_pre_ping": pool_pre_ping}
+    if not is_memory_sqlite:
+        options.update(
+            pool_size=settings.database_pool_size,
+            max_overflow=settings.database_max_overflow,
+            pool_timeout=settings.database_pool_timeout,
+            pool_recycle=settings.database_pool_recycle,
+        )
+
+    logger.info(
+        "Database pool configured",
+        extra={
+            "database_url_backend": backend,
+            "in_memory_sqlite": is_memory_sqlite,
+            **options,
+        },
+    )
+    created_engine = create_async_engine(database_url, echo=False, **options)
+    hold_warn_seconds = settings.database_pool_hold_warn_seconds
+
+    def record_pool_checkout(
+        dbapi_connection: object, connection_record: object, proxy: object
+    ) -> None:
+        connection_record.info["routstr_checked_out_at"] = time.monotonic()  # type: ignore[attr-defined]
+
+    def record_pool_checkin(
+        dbapi_connection: object, connection_record: object
+    ) -> None:
+        checked_out_at = connection_record.info.pop(  # type: ignore[attr-defined]
+            "routstr_checked_out_at", None
+        )
+        if checked_out_at is None:
+            return
+        held_seconds = time.monotonic() - checked_out_at
+        if held_seconds >= hold_warn_seconds:
+            logger.warning(
+                "Database connection held longer than threshold",
+                extra={
+                    "held_seconds": round(held_seconds, 3),
+                    "threshold_seconds": hold_warn_seconds,
+                    "pool_status": created_engine.pool.status(),
+                },
+            )
+
+    event.listen(created_engine.sync_engine, "checkout", record_pool_checkout)
+    event.listen(created_engine.sync_engine, "checkin", record_pool_checkin)
+    return created_engine
+
+
+engine = create_db_engine()
 
 
 class ApiKey(SQLModel, table=True):  # type: ignore
@@ -95,36 +158,194 @@ class ApiKey(SQLModel, table=True):  # type: ignore
 
 
 async def reset_all_reserved_balances(session: AsyncSession) -> None:
-    stmt = update(ApiKey).values(reserved_balance=0, reserved_at=None)
-    await session.exec(stmt)  # type: ignore[call-overload]
+    """Release every active durable reservation during explicit startup reset."""
+    await session.exec(  # type: ignore[call-overload]
+        update(ReservationRelease)
+        .where(col(ReservationRelease.status) == "active")
+        .values(status="released")
+    )
+    await session.exec(  # type: ignore[call-overload]
+        update(ApiKey).values(reserved_balance=0, reserved_at=None)
+    )
     await session.commit()
     logger.info("Reset reserved balances on startup")
 
 
 async def release_stale_reservations(
-    session: AsyncSession, max_age_seconds: int
+    session: AsyncSession,
+    max_age_seconds: int,
+    *,
+    key_hash: str | None = None,
 ) -> int:
-    """Release reservations whose last reserve is older than max_age_seconds.
-    """
+    """Release stale durable reservations without touching newer reservations."""
     cutoff = int(time.time()) - max_age_seconds
-    stmt = (
-        update(ApiKey)
-        .where(col(ApiKey.reserved_balance) > 0)
-        .where(col(ApiKey.reserved_at).is_not(None))
-        .where(col(ApiKey.reserved_at) < cutoff)
-        .values(reserved_balance=0, reserved_at=None)
+    query = (
+        select(ReservationRelease)
+        .where(col(ReservationRelease.status) == "active")
+        .where(col(ReservationRelease.created_at) < cutoff)
     )
-    result = await session.exec(stmt)  # type: ignore[call-overload]
+    if key_hash is not None:
+        query = query.where(
+            or_(
+                col(ReservationRelease.key_hash) == key_hash,
+                col(ReservationRelease.billing_key_hash) == key_hash,
+            )
+        )
+    reservations = (await session.exec(query)).all()
+    released = 0
+
+    for reservation in reservations:
+        transition = await session.exec(  # type: ignore[call-overload]
+            update(ReservationRelease)
+            .where(col(ReservationRelease.id) == reservation.id)
+            .where(col(ReservationRelease.status) == "active")
+            .values(status="released")
+        )
+        if transition.rowcount != 1:
+            continue
+
+        values = {
+            "reserved_balance": col(ApiKey.reserved_balance)
+            - reservation.reserved_msats,
+            "reserved_at": case(
+                (
+                    col(ApiKey.reserved_balance) - reservation.reserved_msats > 0,
+                    col(ApiKey.reserved_at),
+                ),
+                else_=None,
+            ),
+        }
+        parent_result = await session.exec(  # type: ignore[call-overload]
+            update(ApiKey)
+            .where(col(ApiKey.hashed_key) == reservation.billing_key_hash)
+            .where(col(ApiKey.reserved_balance) >= reservation.reserved_msats)
+            .values(**values)
+        )
+        if parent_result.rowcount != 1:
+            await session.rollback()
+            return 0
+
+        if reservation.billing_key_hash != reservation.key_hash:
+            child_result = await session.exec(  # type: ignore[call-overload]
+                update(ApiKey)
+                .where(col(ApiKey.hashed_key) == reservation.key_hash)
+                .where(col(ApiKey.reserved_balance) >= reservation.reserved_msats)
+                .values(**values)
+            )
+            if child_result.rowcount != 1:
+                await session.rollback()
+                return 0
+        released += 1
+
+    # Rolling upgrades can leave aggregate reservations created before durable
+    # reservation rows existed. Release only stale aggregates that have no active
+    # durable owner; targeted refund cleanup also heals legacy NULL timestamps.
+    legacy_query = select(ApiKey).where(col(ApiKey.reserved_balance) > 0)
+    if key_hash is None:
+        legacy_query = legacy_query.where(col(ApiKey.reserved_at).is_not(None)).where(
+            col(ApiKey.reserved_at) < cutoff
+        )
+    else:
+        legacy_query = legacy_query.where(
+            or_(
+                col(ApiKey.hashed_key) == key_hash,
+                col(ApiKey.parent_key_hash) == key_hash,
+            )
+        ).where(
+            or_(col(ApiKey.reserved_at).is_(None), col(ApiKey.reserved_at) < cutoff)
+        )
+
+    for legacy_key in (await session.exec(legacy_query)).all():
+        active_owner = (
+            await session.exec(
+                select(ReservationRelease.id)
+                .where(col(ReservationRelease.status) == "active")
+                .where(
+                    or_(
+                        col(ReservationRelease.key_hash) == legacy_key.hashed_key,
+                        col(ReservationRelease.billing_key_hash)
+                        == legacy_key.hashed_key,
+                    )
+                )
+                .limit(1)
+            )
+        ).first()
+        if active_owner is not None:
+            continue
+        legacy_key.reserved_balance = 0
+        legacy_key.reserved_at = None
+        session.add(legacy_key)
+        released += 1
+
     await session.commit()
-    released = int(result.rowcount or 0)
     if released:
         logger.warning(
-            "Released stale balance reservations",
-            extra={"released_keys": released, "max_age_seconds": max_age_seconds},
+            "Released stale reservations",
+            extra={
+                "released_reservations": released,
+                "max_age_seconds": max_age_seconds,
+            },
         )
     return released
 
 
+async def prune_dead_api_keys(session: AsyncSession, min_age_seconds: int) -> int:
+    """Delete dead parentless API keys; return the count removed.
+
+    Dead = 0 balance/reservation/spend/requests, older than the grace period,
+    no parent, no children, no pending invoice. Cashu rows are unlinked (not
+    deleted) first to keep the audit trail.
+    """
+    cutoff = int(time.time()) - min_age_seconds
+
+    child = aliased(ApiKey)
+    has_children = (
+        select(child.hashed_key).where(
+            col(child.parent_key_hash) == col(ApiKey.hashed_key)
+        )
+    ).exists()
+    pending_invoice = (
+        select(LightningInvoice.id)
+        .where(col(LightningInvoice.api_key_hash) == col(ApiKey.hashed_key))
+        .where(col(LightningInvoice.status) == "pending")
+    ).exists()
+
+    eligible_hashes = (
+        select(ApiKey.hashed_key)
+        .where(col(ApiKey.balance) == 0)
+        .where(col(ApiKey.reserved_balance) == 0)
+        .where(col(ApiKey.total_spent) == 0)
+        .where(col(ApiKey.total_requests) == 0)
+        .where(col(ApiKey.parent_key_hash).is_(None))
+        .where(
+            (col(ApiKey.created_at).is_(None)) | (col(ApiKey.created_at) < cutoff)
+        )
+        .where(~pending_invoice)
+        .where(~has_children)
+    )
+
+    # Unlink transactions rather than cascade-deleting them, so the financial
+    # audit trail survives. The eligibility predicate is re-evaluated inside both
+    # statements so a key that gained balance mid-run is left untouched.
+    await session.exec(  # type: ignore[call-overload]
+        update(CashuTransaction)
+        .where(col(CashuTransaction.api_key_hashed_key).in_(eligible_hashes))
+        .values(api_key_hashed_key=None)
+    )
+
+    result = await session.exec(  # type: ignore[call-overload]
+        delete(ApiKey).where(col(ApiKey.hashed_key).in_(eligible_hashes))
+    )
+    await session.commit()
+
+    pruned = int(result.rowcount or 0)
+    logger.info(
+        "Pruned dead API keys",
+        extra={"pruned_keys": pruned, "min_age_seconds": min_age_seconds},
+    )
+    return pruned
+
+
 class ModelRow(SQLModel, table=True):  # type: ignore
     __tablename__ = "models"
     id: str = Field(primary_key=True)
@@ -161,7 +382,8 @@ class LightningInvoice(SQLModel, table=True):  # type: ignore
     description: str = Field(description="Invoice description")
     payment_hash: str = Field(description="Payment hash for tracking", unique=True)
     status: str = Field(
-        default="pending", description="pending, paid, expired, cancelled"
+        default="pending",
+        description="pending, paid, expired, cancelled, reconciliation_required",
     )
     api_key_hash: str | None = Field(
         default=None, description="Associated API key hash for topup operations"
@@ -206,6 +428,10 @@ class CashuTransaction(SQLModel, table=True):  # type: ignore
     )
     collected: bool = Field(default=False)
     swept: bool = Field(default=False)
+    sweep_started_at: int | None = Field(
+        default=None,
+        description="Unix timestamp for a recoverable refund-sweep claim",
+    )
     source: str = Field(
         default="x-cashu",
         description="Payment source: x-cashu or apikey",
@@ -229,10 +455,13 @@ async def store_cashu_transaction(
     created_at: int | None = None,
     source: str = "x-cashu",
     api_key_hashed_key: str | None = None,
-) -> None:
+    transaction_id: str | None = None,
+    log_failure: bool = True,
+) -> bool:
     try:
         async with create_session() as session:
             tx = CashuTransaction(
+                id=transaction_id or uuid.uuid4().hex,
                 token=token,
                 amount=amount,
                 unit=unit,
@@ -246,11 +475,93 @@ async def store_cashu_transaction(
             )
             session.add(tx)
             await session.commit()
-    except Exception as e:
-        logger.warning(
-            f"Failed to store cashu transaction: {e} (type={typ})",
-            extra={"error": str(e), "type": typ},
-        )
+    except Exception:
+        if log_failure:
+            logger.critical(
+                "Failed to store Cashu transaction",
+                extra={"type": typ, "request_id": request_id, "source": source},
+                exc_info=True,
+            )
+        raise
+    return True
+
+
+async def _cashu_transaction_exists(transaction_id: str) -> bool:
+    async with create_session() as session:
+        return await session.get(CashuTransaction, transaction_id) is not None
+
+
+async def store_cashu_transaction_with_retry(
+    token: str,
+    amount: int,
+    unit: str,
+    mint_url: str | None = None,
+    typ: str = "out",
+    request_id: str | None = None,
+    collected: bool = False,
+    created_at: int | None = None,
+    source: str = "x-cashu",
+    api_key_hashed_key: str | None = None,
+    max_attempts: int = 3,
+) -> bool:
+    """Retry a critical Cashu transaction write with bounded backoff."""
+    transaction_id = hashlib.sha256(f"{typ}\0{token}".encode()).hexdigest()
+    last_error: Exception | None = None
+    for attempt in range(1, max_attempts + 1):
+        try:
+            return await store_cashu_transaction(
+                token=token,
+                amount=amount,
+                unit=unit,
+                mint_url=mint_url,
+                typ=typ,
+                request_id=request_id,
+                collected=collected,
+                created_at=created_at,
+                source=source,
+                api_key_hashed_key=api_key_hashed_key,
+                transaction_id=transaction_id,
+                log_failure=False,
+            )
+        except IntegrityError as error:
+            try:
+                if await _cashu_transaction_exists(transaction_id):
+                    return True
+            except Exception as lookup_error:
+                last_error = lookup_error
+            else:
+                last_error = error
+        except Exception as error:
+            last_error = error
+
+        if last_error is not None:
+            if attempt == max_attempts:
+                break
+            delay = 0.25 * (2 ** (attempt - 1))
+            logger.warning(
+                "Cashu transaction storage failed; retrying",
+                extra={
+                    "type": typ,
+                    "request_id": request_id,
+                    "attempt": attempt,
+                    "max_attempts": max_attempts,
+                    "retry_delay_seconds": delay,
+                },
+            )
+            await asyncio.sleep(delay)
+
+    logger.critical(
+        "Cashu transaction storage failed after bounded retries",
+        extra={
+            "type": typ,
+            "request_id": request_id,
+            "attempts": max_attempts,
+            "error": str(last_error),
+        },
+    )
+    if last_error is None:
+        raise RuntimeError("Cashu transaction storage failed without an exception")
+    raise last_error
 
 
 class UpstreamProviderRow(SQLModel, table=True):  # type: ignore
@@ -261,6 +572,12 @@ class UpstreamProviderRow(SQLModel, table=True):  # type: ignore
         ),
     )
     id: int | None = Field(default=None, primary_key=True)
+    slug: str | None = Field(
+        default=None,
+        unique=True,
+        index=True,
+        description="Stable external slug used for updates via API key.",
+    )
     provider_type: str = Field(
         description="Provider type: custom, openai, anthropic, azure, openrouter, etc."
     )
@@ -282,12 +599,64 @@ class UpstreamProviderRow(SQLModel, table=True):  # type: ignore
     )
 
 
+class ReservationRelease(SQLModel, table=True):  # type: ignore
+    __tablename__ = "reservation_releases"
+    __table_args__ = (
+        Index("ix_reservation_releases_status_created_at", "status", "created_at"),
+    )
+
+    id: str = Field(primary_key=True)
+    key_hash: str = Field(index=True)
+    billing_key_hash: str = Field(index=True)
+    reserved_msats: int
+    status: str = Field(default="active")
+    created_at: int = Field(default_factory=lambda: int(time.time()))
+
+
 class RoutstrFee(SQLModel, table=True):  # type: ignore
     __tablename__ = "routstr_fees"
     id: int = Field(default=1, primary_key=True)
     accumulated_msats: int = Field(default=0)
     total_paid_msats: int = Field(default=0)
     last_paid_at: int | None = Field(default=None)
+    payout_in_progress_msats: int = Field(default=0)
+    payout_started_at: int | None = Field(default=None)
+
+
+class NsecState(str, Enum):
+    """Ownership state of the node's nsec — an explicit 3-state machine.
+
+    The single ``encrypted_nsec`` column cannot distinguish "never migrated" from
+    "intentionally cleared" (both leave it empty), which let a cleared identity be
+    resurrected from a stale legacy ``NSEC``. This names the three states so the
+    bootstrap branches on ownership rather than inferring it:
+
+    * ``legacy`` — the vault has not taken ownership; a plaintext ``NSEC`` (env or
+      old settings blob) may still exist and should be migrated in once.
+    * ``encrypted`` — the vault owns a ciphertext; decrypt it, never re-read env.
+    * ``cleared`` — the vault owns it but the operator emptied it; stay empty,
+      never re-import from a stale legacy copy.
+    """
+
+    legacy = "legacy"
+    encrypted = "encrypted"
+    cleared = "cleared"
+
+
+class Secret(SQLModel, table=True):  # type: ignore
+    """Node-level secrets, stored encrypted/hashed at rest (singleton, id=1).
+
+    The asymmetric column names document the encoding: ``_hash`` is one-way
+    (scrypt, verify only) while ``encrypted_`` is reversible (Fernet). Per-provider
+    upstream keys live on ``upstream_providers``, not here. See ``routstr.core.vault``.
+    """
+
+    __tablename__ = "secrets"
+    id: int = Field(default=1, primary_key=True)
+    admin_password_hash: str | None = Field(default=None)
+    encrypted_nsec: str | None = Field(default=None)
+    nsec_state: NsecState = Field(default=NsecState.legacy)
+    updated_at: int | None = Field(default=None)
 
 
 class CliToken(SQLModel, table=True):  # type: ignore
@@ -328,28 +697,136 @@ async def get_routstr_fee(session: AsyncSession) -> RoutstrFee:
     return fee
 
 
-async def reset_routstr_fee(session: AsyncSession, paid_msats: int) -> None:
+async def get_secret(session: AsyncSession) -> Secret:
+    secret = await session.get(Secret, 1)
+    if secret is None:
+        secret = Secret(id=1)
+        session.add(secret)
+        try:
+            await session.commit()
+        except IntegrityError:
+            # Another worker created the singleton row between our read and
+            # insert (multiple workers booting against one shared DB). Roll back
+            # and read the row they committed instead of failing startup.
+            await session.rollback()
+            secret = await session.get(Secret, 1)
+            if secret is None:
+                raise
+            return secret
+        await session.refresh(secret)
+    return secret
+
+
+async def set_admin_password(session: AsyncSession, password: str) -> None:
+    """Store the admin password as a one-way hash on the Secret singleton."""
+    from .vault import hash_password
+
+    secret = await get_secret(session)
+    secret.admin_password_hash = hash_password(password)
+    secret.updated_at = int(time.time())
+    session.add(secret)
+    await session.commit()
+
+
+async def set_nsec(session: AsyncSession, nsec: str) -> None:
+    """Store the node's nsec, Fernet-encrypted, on the Secret singleton.
+
+    An empty string clears it (the node then holds no Nostr identity and signs
+    no events). Either way the vault now owns the nsec, so the state moves off
+    ``legacy``: a cleared identity (``cleared``) must not be resurrected from a
+    stale legacy ``NSEC`` on the next boot.
+    """
+    from .vault import encrypt
+
+    secret = await get_secret(session)
+    secret.encrypted_nsec = encrypt(nsec) if nsec else None
+    secret.nsec_state = NsecState.encrypted if nsec else NsecState.cleared
+    secret.updated_at = int(time.time())
+    session.add(secret)
+    await session.commit()
+
+
+async def reset_routstr_fee(session: AsyncSession, paid_msats: int) -> bool:
+    """Checkpoint a fee payout before making the external payment."""
     stmt = (
         update(RoutstrFee)
         .where(col(RoutstrFee.id) == 1)
+        .where(col(RoutstrFee.payout_in_progress_msats) == 0)
+        .where(col(RoutstrFee.accumulated_msats) >= paid_msats)
         .values(
             accumulated_msats=RoutstrFee.accumulated_msats - paid_msats,
+            payout_in_progress_msats=paid_msats,
+            payout_started_at=int(time.time()),
+        )
+    )
+    result = await session.exec(stmt)  # type: ignore[call-overload]
+    await session.commit()
+    return result.rowcount == 1
+
+
+async def complete_routstr_fee_payout(
+    session: AsyncSession, paid_msats: int
+) -> bool:
+    """Mark a checkpointed payout complete after the external payment succeeds."""
+    stmt = (
+        update(RoutstrFee)
+        .where(col(RoutstrFee.id) == 1)
+        .where(col(RoutstrFee.payout_in_progress_msats) == paid_msats)
+        .values(
+            payout_in_progress_msats=0,
+            payout_started_at=None,
             total_paid_msats=RoutstrFee.total_paid_msats + paid_msats,
             last_paid_at=int(time.time()),
         )
     )
-    await session.exec(stmt)  # type: ignore[call-overload]
+    result = await session.exec(stmt)  # type: ignore[call-overload]
     await session.commit()
+    return result.rowcount == 1
 
 
-async def balances_for_mint_and_unit(
+async def total_user_liability(db_session: AsyncSession) -> int:
+    """Return all outstanding API-key balances in millisatoshis."""
+    result = await db_session.exec(select(func.sum(ApiKey.balance)))
+    return int(result.one() or 0)
+
+
+async def balance_for_mint_and_unit(
     db_session: AsyncSession, mint_url: str, unit: str
 ) -> int:
-    query = select(func.sum(ApiKey.balance)).where(
-        ApiKey.refund_mint_url == mint_url, ApiKey.refund_currency == unit
+    """Return the user liability for one mint and unit in millisatoshis."""
+    result = await db_session.exec(
+        select(func.sum(ApiKey.balance)).where(
+            col(ApiKey.refund_mint_url) == mint_url,
+            col(ApiKey.refund_currency) == unit,
+        )
+    )
+    return int(result.one() or 0)
+
+
+async def balances_by_mint_and_unit(
+    db_session: AsyncSession, mint_urls: list[str], units: list[str]
+) -> dict[tuple[str, str], int]:
+    """Return requested user liabilities grouped by mint and unit."""
+    if not mint_urls or not units:
+        return {}
+    query = (
+        select(
+            col(ApiKey.refund_mint_url),
+            col(ApiKey.refund_currency),
+            func.sum(ApiKey.balance),
+        )
+        .where(
+            col(ApiKey.refund_mint_url).in_(mint_urls),
+            col(ApiKey.refund_currency).in_(units),
+        )
+        .group_by(col(ApiKey.refund_mint_url), col(ApiKey.refund_currency))
     )
     result = await db_session.exec(query)
-    return result.one() or 0
+    return {
+        (mint_url, unit): int(balance or 0)
+        for mint_url, unit, balance in result.all()
+        if mint_url is not None and unit is not None
+    }
 
 
 async def init_db() -> None:

routstr/core/exceptions.py

@@ -7,11 +7,26 @@ logger = get_logger(__name__)
 
 
 class UpstreamError(Exception):
-    """Exception raised when an upstream provider fails."""
+    """Exception raised when an upstream provider fails.
 
-    def __init__(self, message: str, status_code: int = 502):
+    ``code`` carries a stable, machine-readable classification (e.g.
+    ``UPSTREAM_RATE_LIMIT``) so callers can distinguish failure kinds without
+    string-matching the message. ``details`` holds optional structured,
+    redaction-safe context. Both default to ``None`` for backwards
+    compatibility.
+    """
+
+    def __init__(
+        self,
+        message: str,
+        status_code: int = 502,
+        code: str | None = None,
+        details: dict[str, object] | None = None,
+    ):
         self.message = message
         self.status_code = status_code
+        self.code = code
+        self.details = details
         super().__init__(message)
 
 

routstr/core/logging.py

@@ -51,6 +51,8 @@ from pythonjsonlogger import jsonlogger
 from rich.console import Console
 from rich.logging import RichHandler
 
+from .redaction import redact_obj, redact_org_ids
+
 # Only use RichHandler when stdout is a real TTY. In non-TTY contexts
 # (docker logs, pipes, CI) Rich pads every line to width and wraps long
 # records, producing visually-empty trailing whitespace and split records.
@@ -180,6 +182,37 @@ class RequestIdFilter(logging.Filter):
         return True
 
 
+# Standard ``LogRecord`` attributes that are never user-supplied ``extra``
+# fields; skipped when redacting structured extras (``msg``/``message`` are
+# handled separately above).
+_NON_EXTRA_RECORD_ATTRS = frozenset(
+    {
+        "name",
+        "msg",
+        "args",
+        "levelname",
+        "levelno",
+        "pathname",
+        "filename",
+        "module",
+        "exc_info",
+        "exc_text",
+        "stack_info",
+        "lineno",
+        "funcName",
+        "created",
+        "msecs",
+        "relativeCreated",
+        "thread",
+        "threadName",
+        "processName",
+        "process",
+        "taskName",
+        "message",
+    }
+)
+
+
 class SecurityFilter(logging.Filter):
     """Filter to remove sensitive information from logs."""
 
@@ -203,6 +236,7 @@ class SecurityFilter(logging.Filter):
         """Filter out sensitive information from log records."""
         try:
             message = record.getMessage()
+            message = redact_org_ids(message)
             standalone_patterns = [
                 r"Bearer\s+([a-zA-Z0-9_\-\.]{10,})",  # Bearer token (must be 10 characters or more to reduce false-positives)
                 r"cashu[A-Z]+([a-zA-Z0-9_\-\.=/+]+)",  # Cashu tokens
@@ -224,6 +258,16 @@ class SecurityFilter(logging.Filter):
             record.msg = message
             record.args = ()
 
+            # Structured `extra={...}` fields are emitted by the JSON formatter
+            # straight from the record dict and never pass through the message
+            # formatting above. Redact organization IDs from any string-valued
+            # extra so they cannot leak via structured logs.
+            for attr, value in list(record.__dict__.items()):
+                if attr in _NON_EXTRA_RECORD_ATTRS:
+                    continue
+                if isinstance(value, (str, dict, list, tuple)):
+                    record.__dict__[attr] = redact_obj(value)
+
         except Exception:
             pass
 

routstr/core/main.py

@@ -11,7 +11,11 @@ from starlette.exceptions import HTTPException
 from starlette.responses import Response as StarletteResponse
 from starlette.types import Scope
 
-from ..auth import periodic_key_reset, periodic_stale_reservation_sweep
+from ..auth import (
+    periodic_dead_key_prune,
+    periodic_key_reset,
+    periodic_stale_reservation_sweep,
+)
 from ..balance import balance_router, deprecated_wallet_router
 from ..lightning import lightning_router, periodic_invoice_watcher
 from ..nostr import (
@@ -24,6 +28,7 @@ from ..payment.models import models_router, update_sats_pricing
 from ..payment.price import update_prices_periodically
 from ..proxy import initialize_upstreams, proxy_router, refresh_model_maps_periodically
 from ..upstream.auto_topup import periodic_auto_topup
+from ..upstream.deepseek_v4_pricing_shim import register_deepseek_v4_pricing
 from ..upstream.litellm_routing import configure_litellm
 from ..wallet import periodic_payout, periodic_refund_sweep, periodic_routstr_fee_payout
 from .admin import admin_router
@@ -32,7 +37,7 @@ from .exceptions import general_exception_handler, http_exception_handler
 from .logging import get_logger, setup_logging
 from .middleware import LoggingMiddleware
 from .not_found import _NOT_FOUND_HTML, not_found_catch_all  # noqa: F401
-from .settings import SettingsService
+from .settings import SettingsService, bootstrap_secrets
 from .settings import settings as global_settings
 from .version import __version__
 
@@ -55,6 +60,7 @@ async def lifespan(_: FastAPI) -> AsyncGenerator[None, None]:
     model_maps_refresh_task = None
     key_reset_task = None
     stale_reservation_task = None
+    dead_key_prune_task = None
     auto_topup_task = None
     refund_sweep_task = None
     routstr_fee_task = None
@@ -65,6 +71,11 @@ async def lifespan(_: FastAPI) -> AsyncGenerator[None, None]:
         # debug logging) before any upstream provider dispatches a request.
         configure_litellm()
 
+        # TEMPORARY: backfill DeepSeek V4 pricing missing from litellm's cost
+        # map (BerriAI/litellm#30430). Remove this call and
+        # deepseek_v4_pricing_shim.py once litellm ships these models.
+        register_deepseek_v4_pricing()
+
         # Run database migrations on startup
         run_migrations()
 
@@ -74,17 +85,20 @@ async def lifespan(_: FastAPI) -> AsyncGenerator[None, None]:
 
         # Initialize application settings (env -> computed -> DB precedence)
         async with create_session() as session:
+            # Move secrets into the encrypted/hashed store and decrypt the nsec
+            # into the in-memory settings BEFORE initializing settings: the
+            # initialize step strips secrets from the persisted blob, so legacy
+            # plaintext (env or old blob) must be migrated into the Secret store
+            # first or the only copy of a blob-only secret would be lost.
+            # Generates and logs an admin password on a fresh node; fails fast if
+            # a stored secret can't be decrypted.
+            await bootstrap_secrets(session)
             s = await SettingsService.initialize(session)
             if s.reset_reserved_balance_on_startup:
                 from .db import reset_all_reserved_balances
 
                 await reset_all_reserved_balances(session)
 
-        if not s.admin_password:
-            logger.warning(
-                f"Admin password is not set. Visit {s.http_url or 'http://localhost:8000'}/admin to set the password."
-            )
-
         # Apply app metadata from settings
         try:
             app.title = s.name
@@ -126,6 +140,7 @@ async def lifespan(_: FastAPI) -> AsyncGenerator[None, None]:
         stale_reservation_task = asyncio.create_task(
             periodic_stale_reservation_sweep()
         )
+        dead_key_prune_task = asyncio.create_task(periodic_dead_key_prune())
         auto_topup_task = asyncio.create_task(periodic_auto_topup())
         refund_sweep_task = asyncio.create_task(periodic_refund_sweep())
         routstr_fee_task = asyncio.create_task(periodic_routstr_fee_payout())
@@ -165,6 +180,8 @@ async def lifespan(_: FastAPI) -> AsyncGenerator[None, None]:
             key_reset_task.cancel()
         if stale_reservation_task is not None:
             stale_reservation_task.cancel()
+        if dead_key_prune_task is not None:
+            dead_key_prune_task.cancel()
         if auto_topup_task is not None:
             auto_topup_task.cancel()
         if refund_sweep_task is not None:
@@ -196,6 +213,8 @@ async def lifespan(_: FastAPI) -> AsyncGenerator[None, None]:
                 tasks_to_wait.append(key_reset_task)
             if stale_reservation_task is not None:
                 tasks_to_wait.append(stale_reservation_task)
+            if dead_key_prune_task is not None:
+                tasks_to_wait.append(dead_key_prune_task)
             if auto_topup_task is not None:
                 tasks_to_wait.append(auto_topup_task)
             if refund_sweep_task is not None:
@@ -241,7 +260,20 @@ app.add_middleware(
     allow_credentials=True,
     allow_methods=["*"],
     allow_headers=["*"],
-    expose_headers=["x-routstr-request-id", "x-cashu"],
+    expose_headers=[
+        "x-routstr-request-id",
+        "x-cashu",
+        "x-routstr-cost-msats",
+        "x-routstr-cost-usd",
+        "x-routstr-input-cost-msats",
+        "x-routstr-output-cost-msats",
+        # EHBP (Tinfoil) protocol headers must be exposed so browser clients
+        # can detect and decrypt encrypted responses. Without these, the
+        # browser hides them via CORS and the SDK treats the response as a
+        # plaintext proxy error, returning raw ciphertext.
+        "Ehbp-Response-Nonce",
+        "Ehbp-Encapsulated-Key",
+    ],
 )
 
 # Add logging middleware
@@ -363,7 +395,10 @@ if UI_DIST_PATH.exists() and UI_DIST_PATH.is_dir():
 
 else:
     logger.warning(
-        f"UI dist directory not found at {UI_DIST_PATH}, skipping static file serving"
+        "UI dist directory not found at %s; serving API only. Run `make ui-build` "
+        "to build the static UI served from here, or `make ui-dev` for the Next.js "
+        "dev server with hot reload on :3000 (it targets this backend on :8000).",
+        UI_DIST_PATH,
     )
 
     @app.get("/", include_in_schema=False)

routstr/core/provider_slugs.py

@@ -0,0 +1,65 @@
+from __future__ import annotations
+
+import re
+from itertools import count
+from typing import Collection
+
+from sqlmodel import select
+from sqlmodel.ext.asyncio.session import AsyncSession
+
+from .db import UpstreamProviderRow
+
+_SLUG_BASE_PATTERN = re.compile(r"[^a-z0-9]+")
+_MAX_SLUG_LENGTH = 64
+
+
+def provider_slug_base(provider_type: str) -> str:
+    """Return a deterministic slug base for a provider type."""
+    base = _SLUG_BASE_PATTERN.sub("-", provider_type.lower()).strip("-")
+    if not base:
+        base = "provider"
+    elif base.isdigit():
+        base = f"provider-{base}"
+    elif len(base) < 3:
+        base = f"{base}-provider"
+
+    if len(base) > _MAX_SLUG_LENGTH:
+        base = base[:_MAX_SLUG_LENGTH].rstrip("-") or "provider"
+    return base
+
+
+def provider_slug_candidate(base: str, suffix_number: int) -> str:
+    if suffix_number == 1:
+        return base
+
+    suffix = f"-{suffix_number}"
+    max_base_length = _MAX_SLUG_LENGTH - len(suffix)
+    return f"{base[:max_base_length].rstrip('-')}{suffix}"
+
+
+async def allocate_unique_provider_slug(
+    session: AsyncSession,
+    provider_type: str,
+    reserved_slugs: Collection[str] = (),
+) -> str:
+    """Allocate a stable, deterministic provider slug.
+
+    The first provider of a type gets ``openai``; later collisions get
+    ``openai-2``, ``openai-3``, etc. ``reserved_slugs`` covers rows staged in
+    memory but not flushed yet, such as settings/env seeding.
+    """
+    base = provider_slug_base(provider_type)
+    reserved = {slug.lower() for slug in reserved_slugs}
+
+    for suffix_number in count(1):
+        candidate = provider_slug_candidate(base, suffix_number)
+        if candidate in reserved:
+            continue
+
+        result = await session.exec(
+            select(UpstreamProviderRow).where(UpstreamProviderRow.slug == candidate)
+        )
+        if result.first() is None:
+            return candidate
+
+    raise RuntimeError("unreachable")

routstr/core/redaction.py

@@ -0,0 +1,51 @@
+"""Redaction helpers for sensitive provider identifiers.
+
+Single source of truth for stripping account-scoped identifiers (e.g. OpenAI
+organization IDs) from any text before it is logged, returned to a caller, or
+written to an audit entry.
+"""
+
+from __future__ import annotations
+
+import re
+from typing import Any
+
+# OpenAI-style organization identifiers look like ``org-<base62>``. Require at
+# least 6 trailing chars so the already-redacted literal ``org-[REDACTED]`` is
+# never re-matched (``[`` is not in the character class).
+_ORG_ID_PATTERN = re.compile(r"\borg-[A-Za-z0-9]{6,}\b")
+
+ORG_ID_PLACEHOLDER = "org-[REDACTED]"
+
+
+def redact_org_ids(text: str) -> str:
+    """Replace OpenAI-style organization IDs with ``org-[REDACTED]``.
+
+    Args:
+        text: Arbitrary text that may embed an ``org-*`` identifier.
+
+    Returns:
+        The text with every organization ID replaced. Non-string input is
+        returned unchanged after coercion to ``str``.
+    """
+    if not text:
+        return text
+    return _ORG_ID_PATTERN.sub(ORG_ID_PLACEHOLDER, text)
+
+
+def redact_obj(obj: Any) -> Any:
+    """Recursively redact organization IDs in arbitrary nested structures.
+
+    Strings are redacted in place; dicts and lists/tuples are walked so that
+    identifiers nested inside structured payloads (e.g. log ``extra`` fields or
+    error ``details``) are also stripped. Other types are returned unchanged.
+    """
+    if isinstance(obj, str):
+        return redact_org_ids(obj)
+    if isinstance(obj, dict):
+        return {key: redact_obj(value) for key, value in obj.items()}
+    if isinstance(obj, list):
+        return [redact_obj(value) for value in obj]
+    if isinstance(obj, tuple):
+        return tuple(redact_obj(value) for value in obj)
+    return obj

routstr/core/settings.py

@@ -3,6 +3,8 @@ from __future__ import annotations
 import asyncio
 import json
 import os
+import secrets
+import time
 from datetime import datetime, timezone
 from typing import Any
 
@@ -26,7 +28,6 @@ class Settings(BaseSettings):
     # Core
     upstream_base_url: str = Field(default="", env="UPSTREAM_BASE_URL")
     upstream_api_key: str = Field(default="", env="UPSTREAM_API_KEY")
-    admin_password: str = Field(default="", env="ADMIN_PASSWORD")
 
     # Node info
     name: str = Field(default="ARoutstrNode", env="NAME")
@@ -40,6 +41,9 @@ class Settings(BaseSettings):
     receive_ln_address: str = Field(default="", env="RECEIVE_LN_ADDRESS")
     primary_mint: str = Field(default="", env="PRIMARY_MINT_URL")
     primary_mint_unit: str = Field(default="sat", env="PRIMARY_MINT_UNIT")
+    mint_operation_concurrency: int = Field(
+        default=4, ge=1, env="MINT_OPERATION_CONCURRENCY"
+    )
 
     # Lightning payout configuration
     # Minimum available balance (in satoshis) before profit is paid out over
@@ -73,6 +77,14 @@ class Settings(BaseSettings):
     stale_reservation_timeout_seconds: int = Field(
         default=300, env="STALE_RESERVATION_TIMEOUT_SECONDS"
     )
+    # Background prune of dead (zero balance, never used) API keys.
+    # Interval 0 disables it; min-age is a grace period (default 1 week).
+    dead_key_prune_interval_seconds: int = Field(
+        default=3600, env="DEAD_KEY_PRUNE_INTERVAL_SECONDS"
+    )
+    dead_key_min_age_seconds: int = Field(
+        default=604_800, env="DEAD_KEY_MIN_AGE_SECONDS"
+    )
 
     # Network
     cors_origins: list[str] = Field(default_factory=lambda: ["*"], env="CORS_ORIGINS")
@@ -90,6 +102,24 @@ class Settings(BaseSettings):
     enable_models_refresh: bool = Field(default=True, env="ENABLE_MODELS_REFRESH")
     refund_cache_ttl_seconds: int = Field(default=3600, env="REFUND_CACHE_TTL_SECONDS")
     refund_sweep_ttl_seconds: int = Field(default=604800, env="REFUND_SWEEP_TTL_SECONDS")
+    refund_sweep_claim_timeout_seconds: int = Field(
+        default=900, gt=0, env="REFUND_SWEEP_CLAIM_TIMEOUT_SECONDS"
+    )
+
+    # Database connection-pool controls (advanced). Capacity defaults match
+    # SQLAlchemy's established queue-pool behavior. Pre-ping is enabled by the
+    # engine factory for networked backends; SQLite can explicitly opt in.
+    # These fields are env-only below.
+    database_pool_size: int = Field(default=5, ge=1, env="DATABASE_POOL_SIZE")
+    database_max_overflow: int = Field(default=10, ge=0, env="DATABASE_MAX_OVERFLOW")
+    database_pool_timeout: float = Field(
+        default=30.0, gt=0, env="DATABASE_POOL_TIMEOUT"
+    )
+    database_pool_recycle: int = Field(default=1800, ge=0, env="DATABASE_POOL_RECYCLE")
+    database_pool_pre_ping: bool = Field(default=False, env="DATABASE_POOL_PRE_PING")
+    database_pool_hold_warn_seconds: float = Field(
+        default=10.0, gt=0, env="DATABASE_POOL_HOLD_WARN_SECONDS"
+    )
 
     # Logging
     log_level: str = Field(default="INFO", env="LOG_LEVEL")
@@ -124,10 +154,92 @@ def _normalize_settings_data(data: dict[str, Any]) -> dict[str, Any]:
     return normalized
 
 
+# Secrets are credentials, not config: they live in the encrypted/hashed Secret
+# store (and decrypted in-memory for runtime use), never in the persisted
+# settings blob. ``admin_password`` is gone from the model entirely; ``nsec``
+# remains a live field but is stripped from every blob write so it is never
+# written back to plaintext. ``upstream_api_key`` is intentionally *not* here:
+# it has no encrypted home yet (it is node-scoped today but really belongs on a
+# provider), so stripping it would lose it on the next restart. It stays in the
+# blob as before; encrypting it is follow-up work. See ``bootstrap_secrets`` and
+# ``routstr.core.vault``.
+SECRET_FIELDS = frozenset({"admin_password", "nsec"})
+
+# Infrastructure the node needs *before* it can open a DB session — so it can
+# never be configured from the DB (chicken-and-egg) and stays env-only. Unlike
+# secrets (owned by bootstrap), these are excluded so the DB settings blob can
+# neither store nor shadow them; env is always authoritative.
+ENV_ONLY_FIELDS = frozenset(
+    {
+        "database_pool_size",
+        "database_max_overflow",
+        "database_pool_timeout",
+        "database_pool_recycle",
+        "database_pool_pre_ping",
+        "database_pool_hold_warn_seconds",
+    }
+)
+
+_NON_PERSISTED_FIELDS = SECRET_FIELDS | ENV_ONLY_FIELDS
+
+
+def _strip_secret_fields(data: dict[str, Any]) -> dict[str, Any]:
+    """Return a copy of ``data`` without secret or env-only fields.
+
+    Both are kept out of the persisted settings blob: secrets for confidentiality,
+    env-only fields (e.g. DB pool sizing) because they must never be sourced from
+    the database.
+    """
+    return {k: v for k, v in data.items() if k not in _NON_PERSISTED_FIELDS}
+
+
+def _apply_to_live_settings(data: dict[str, Any]) -> None:
+    """Apply ``data`` onto the live ``settings`` for all in-process importers.
+
+    Secrets are owned exclusively by ``bootstrap_secrets``, which runs first and
+    has already decrypted the authoritative nsec into memory (importing any
+    legacy plaintext on the way). Never re-apply secret fields from env/blob
+    here: a non-empty but stale ``NSEC`` env var would otherwise override an nsec
+    the vault has taken ownership of (e.g. after the operator rotates it in the
+    UI), and an empty one would wipe the live value. Skip them entirely.
+    """
+    for k, v in data.items():
+        if k in SECRET_FIELDS:
+            continue
+        setattr(settings, k, v)
+
+
 def _compute_primary_mint(cashu_mints: list[str]) -> str:
     return cashu_mints[0] if cashu_mints else "https://mint.minibits.cash/Bitcoin"
 
 
+def derive_npub_from_nsec(nsec: str) -> str | None:
+    """Derive the npub (bech32) from an nsec or 64-char hex private key, or None.
+
+    Parsing is delegated to :func:`routstr.nostr.listing.nsec_to_keypair`, the
+    single place that knows the nsec/hex formats (and already returns ``None`` on
+    any unusable input); this only bech32-encodes the resulting public key. The
+    contract stays "return None on unusable input", so a bad key never crashes
+    boot.
+    """
+    try:
+        from nostr.key import PublicKey  # type: ignore
+
+        from ..nostr.listing import nsec_to_keypair
+    except ImportError:
+        return None
+
+    keypair = nsec_to_keypair(nsec)
+    if keypair is None:
+        return None
+    _privkey_hex, pubkey_hex = keypair
+
+    try:
+        return PublicKey(bytes.fromhex(pubkey_hex)).bech32()
+    except (ValueError, AttributeError):
+        return None
+
+
 def resolve_bootstrap() -> Settings:
     base = Settings()  # Reads env with custom parse_env_var
     # Back-compat env mapping
@@ -182,23 +294,9 @@ def resolve_bootstrap() -> Settings:
             pass
     # Derive NPUB from NSEC if not provided
     if not base.npub and base.nsec:
-        try:
-            from nostr.key import PrivateKey  # type: ignore
-
-            if base.nsec.startswith("nsec"):
-                pk = PrivateKey.from_nsec(base.nsec)
-            elif len(base.nsec) == 64:
-                pk = PrivateKey(bytes.fromhex(base.nsec))
-            else:
-                pk = None
-            if pk is not None:
-                try:
-                    base.npub = pk.public_key.bech32()
-                except Exception:
-                    # Fallback to hex if bech32 not available
-                    base.npub = pk.public_key.hex()
-        except Exception:
-            pass
+        npub = derive_npub_from_nsec(base.nsec)
+        if npub:
+            base.npub = npub
     if not base.cors_origins:
         base.cors_origins = ["*"]
     if not base.primary_mint:
@@ -248,15 +346,14 @@ class SettingsService:
                     text(
                         "INSERT INTO settings (id, data, updated_at) VALUES (1, :data, :updated_at)"
                     ).bindparams(
-                        data=json.dumps(env_resolved.dict()),
+                        data=json.dumps(_strip_secret_fields(env_resolved.dict())),
                         updated_at=datetime.now(timezone.utc),
                     )
                 )
                 await db_session.commit()
                 cls._current = settings
                 # Update the existing instance in-place for all live importers
-                for k, v in env_resolved.dict().items():
-                    setattr(settings, k, v)
+                _apply_to_live_settings(env_resolved.dict())
                 return cls._current
 
             db_id, db_data, _updated_at = row
@@ -273,7 +370,13 @@ class SettingsService:
             valid_fields = set(env_resolved.dict().keys())
             merged_dict: dict[str, Any] = dict(env_resolved.dict())
             merged_dict.update(
-                {k: v for k, v in db_json.items() if v not in (None, "", [], {}) and k in valid_fields}
+                {
+                    k: v
+                    for k, v in db_json.items()
+                    if v not in (None, "", [], {})
+                    and k in valid_fields
+                    and k not in ENV_ONLY_FIELDS
+                }
             )
             merged_dict = Settings(**merged_dict).dict()
 
@@ -283,20 +386,37 @@ class SettingsService:
                     merged_dict.get("cashu_mints", [])
                 )
 
-            if db_json_raw != merged_dict:
+            # Keep npub consistent with the live nsec. bootstrap_secrets has
+            # already run and holds the single authoritative nsec (decrypted from
+            # the encrypted store, or freshly imported). merged_dict starts from
+            # the env/blob, which may carry a STALE nsec — and therefore a stale
+            # derived npub — after the vault took ownership. Derive from the live
+            # value and OVERRIDE, not just fill: otherwise the node keeps the
+            # vault's private key but announces the old env key's npub (npub is a
+            # pure derivation of nsec, never configured independently of it).
+            if settings.nsec:
+                derived_npub = derive_npub_from_nsec(settings.nsec)
+                if derived_npub:
+                    merged_dict["npub"] = derived_npub
+
+            # Persist without secrets; compare against the stripped target so a
+            # legacy blob that still carries plaintext secrets gets rewritten
+            # (and thereby sunset) even when its non-secret values are unchanged.
+            persisted = _strip_secret_fields(merged_dict)
+            if db_json_raw != persisted:
                 await db_session.exec(  # type: ignore
                     text(
                         "UPDATE settings SET data = :data, updated_at = :updated_at WHERE id = 1"
                     ).bindparams(
-                        data=json.dumps(merged_dict),
+                        data=json.dumps(persisted),
                         updated_at=datetime.now(timezone.utc),
                     )
                 )
                 await db_session.commit()
 
             # Update the existing instance in-place for all live importers
-            for k, v in merged_dict.items():
-                setattr(settings, k, v)
+            # (keeps the decrypted nsec live in memory).
+            _apply_to_live_settings(merged_dict)
             cls._current = settings
             return cls._current
 
@@ -318,13 +438,18 @@ class SettingsService:
                 text(
                     "UPDATE settings SET data = :data, updated_at = :updated_at WHERE id = 1"
                 ).bindparams(
-                    data=json.dumps(candidate.dict()),
+                    data=json.dumps(_strip_secret_fields(candidate.dict())),
                     updated_at=datetime.now(timezone.utc),
                 )
             )
             await db_session.commit()
-            # Update in-place
+            # Update in-place. Env-only fields (e.g. DB pool sizing) are never
+            # applied here: the engine pool is already built at boot from env,
+            # so letting an update mutate the live value would only make it
+            # diverge from the running pool.
             for k, v in candidate.dict().items():
+                if k in ENV_ONLY_FIELDS:
+                    continue
                 setattr(settings, k, v)
             cls._current = settings
             return settings
@@ -347,3 +472,152 @@ class SettingsService:
                     setattr(settings, k, v)
             cls._current = settings
             return settings
+
+
+async def _read_raw_settings_blob(db_session: AsyncSession) -> dict[str, Any]:
+    """Best-effort read of the raw persisted settings JSON (may not exist yet)."""
+    from sqlmodel import text
+
+    try:
+        result = await db_session.exec(  # type: ignore
+            text("SELECT data FROM settings WHERE id = 1")
+        )
+        row = result.first()
+    except Exception:
+        return {}
+    if row is None:
+        return {}
+    (data_str,) = row
+    try:
+        data = json.loads(data_str) if isinstance(data_str, str) else dict(data_str)
+    except Exception:
+        return {}
+    return data if isinstance(data, dict) else {}
+
+
+def _legacy_plaintext(
+    raw_blob: dict[str, Any], env_name: str, blob_key: str
+) -> str | None:
+    """Legacy plaintext for a secret: env first, then the old settings blob."""
+    env_value = os.environ.get(env_name)
+    if env_value:
+        return env_value
+    blob_value = raw_blob.get(blob_key)
+    if isinstance(blob_value, str) and blob_value:
+        return blob_value
+    return None
+
+
+async def bootstrap_secrets(db_session: AsyncSession) -> None:
+    """Move node secrets into the encrypted/hashed Secret store at startup.
+
+    Per secret:
+      * column already set -> use it (the nsec is decrypted into the in-memory
+        ``settings``; a wrong ROUTSTR_SECRET_KEY surfaces as a clear fail-fast).
+      * column empty but legacy plaintext exists (env, or the old settings
+        blob) -> transform it (hash the password / encrypt the nsec) into the
+        column.
+      * nothing (admin password only) -> generate a strong random password,
+        hash it, and log it once with the /admin URL.
+    """
+    from cryptography.fernet import InvalidToken
+    from sqlmodel import col, update
+
+    from . import vault
+    from .db import NsecState, Secret, get_secret
+
+    raw_blob = await _read_raw_settings_blob(db_session)
+    secret = await get_secret(db_session)
+    changed = False
+
+    # Admin password — one-way scrypt hash.
+    if secret.admin_password_hash is None:
+        legacy_password = _legacy_plaintext(
+            raw_blob, "ADMIN_PASSWORD", "admin_password"
+        )
+        if legacy_password:
+            secret.admin_password_hash = vault.hash_password(legacy_password)
+            changed = True
+        else:
+            generated = secrets.token_urlsafe(24)
+            # Claim the empty slot atomically: only the worker whose UPDATE flips
+            # NULL -> hash owns the generated password and announces it. On a
+            # shared DB a racing worker gets rowcount 0, so it neither clobbers
+            # the winner's hash (which the operator may already be using) nor
+            # prints a second password that would never work.
+            claim_stmt = (
+                update(Secret)
+                .where(col(Secret.id) == 1)
+                .where(col(Secret.admin_password_hash).is_(None))
+                .values(
+                    admin_password_hash=vault.hash_password(generated),
+                    updated_at=int(time.time()),
+                )
+            )
+            result = await db_session.exec(claim_stmt)  # type: ignore[call-overload]
+            await db_session.commit()
+            await db_session.refresh(secret)
+            if result.rowcount == 1:
+                admin_url = (settings.http_url or "http://localhost:8000").rstrip("/")
+                # Print to stdout rather than the logger: the operator must see
+                # this once (e.g. `docker compose logs`), but it must not be
+                # persisted into the on-disk log files the logger also writes to.
+                print(
+                    "No admin password set; generated a temporary one (shown "
+                    f"only now): {generated}\nLog in at {admin_url}/admin and "
+                    "change it from the dashboard settings.",
+                    flush=True,
+                )
+
+    # Nostr nsec — reversible Fernet encryption. ``nsec_state`` is the single
+    # source of truth for ownership, so "intentionally cleared" is never
+    # conflated with "never migrated" (the bug the old bool could not encode).
+    if secret.nsec_state == NsecState.encrypted:
+        # The vault owns the identity: decrypt the ciphertext, never re-read
+        # env/blob. A missing ciphertext here means the row is inconsistent (a
+        # failed write or manual edit); fail fast rather than silently dropping
+        # the identity and falling back to a stale legacy copy.
+        if secret.encrypted_nsec is None:
+            raise RuntimeError(
+                "nsec_state is 'encrypted' but no ciphertext is stored; the "
+                "secrets row is inconsistent. Refusing to boot rather than "
+                "silently resurrecting a stale legacy NSEC."
+            )
+        try:
+            settings.nsec = vault.decrypt(secret.encrypted_nsec)
+        except InvalidToken as exc:
+            raise RuntimeError(
+                "Stored nsec cannot be decrypted with the current "
+                "ROUTSTR_SECRET_KEY. The key changed, or this database came from "
+                "another node. Restore the original ROUTSTR_SECRET_KEY to recover."
+            ) from exc
+    elif secret.nsec_state == NsecState.cleared:
+        # The operator emptied the identity via the admin API. A fresh process
+        # has already reloaded a stale ``NSEC`` from env/blob into the live
+        # settings (and may have derived its npub); actively clear both so the
+        # cleared store wins rather than silently resurrecting the old identity.
+        settings.nsec = ""
+        settings.npub = ""
+    else:  # NsecState.legacy — the vault has not taken ownership yet.
+        # Import any legacy plaintext (env, or the old settings blob) exactly
+        # once. Encryption at rest is mandatory, but a missing key is
+        # provisioned, not fatal: vault.encrypt generates and persists a master
+        # key (with a loud one-time operator notice) when none was supplied, so
+        # an upgrading node keeps running. The nsec is never stored in plaintext.
+        legacy_nsec = _legacy_plaintext(raw_blob, "NSEC", "nsec")
+        if legacy_nsec:
+            secret.encrypted_nsec = vault.encrypt(legacy_nsec)
+            secret.nsec_state = NsecState.encrypted
+            settings.nsec = legacy_nsec
+            changed = True
+
+    # Derive npub from whatever nsec we now hold, if not already known.
+    if settings.nsec and not settings.npub:
+        npub = derive_npub_from_nsec(settings.nsec)
+        if npub:
+            settings.npub = npub
+
+    if changed:
+        secret.updated_at = int(time.time())
+        db_session.add(secret)
+        await db_session.commit()

routstr/core/vault.py

@@ -0,0 +1,320 @@
+"""Encrypt/hash/fingerprint helpers for secrets at rest (issue #553).
+
+Thin wrapper over ``cryptography`` so nothing else in the codebase touches
+Fernet/scrypt/HMAC directly:
+
+- :func:`encrypt`/:func:`decrypt` — Fernet symmetric encryption, keyed by the
+  mandatory master key. Ciphertext is self-describing (``fernet:v1:`` prefix) so
+  a value can be told apart from legacy plaintext and so reading it under the
+  wrong key surfaces as a hard error rather than silent corruption.
+- :func:`hash_password`/:func:`verify_password` — salted scrypt hashing. This is
+  *key-independent*: it never reads the master key, so password login and the
+  recovery script keep working even when the key is missing.
+
+Key custody is flexible but encryption is not optional. The key comes from the
+``ROUTSTR_SECRET_KEY`` env var, else a persisted key file
+(``ROUTSTR_SECRET_KEY_FILE``, defaulting beside the SQLite database so it persists
+on the same volume as the data); when neither is set, :func:`encrypt` generates
+one to the key file and prints a one-time notice, so an existing node upgrades
+without breaking instead of refusing to boot. Reading is strict — :func:`decrypt`
+never generates a key (a new key could not match existing ciphertext) and fails
+fast with the generation command when none is configured. A malformed
+``ROUTSTR_SECRET_KEY`` is an operator error and always fails fast.
+"""
+
+import base64
+import hashlib
+import hmac
+import os
+import secrets
+import tempfile
+from pathlib import Path
+
+from cryptography.fernet import Fernet
+from sqlalchemy.engine import make_url
+from sqlalchemy.exc import ArgumentError
+
+_PREFIX = "fernet:v1:"
+_GEN_COMMAND = (
+    'python -c "from cryptography.fernet import Fernet; '
+    'print(Fernet.generate_key().decode())"'
+)
+
+# Where an auto-generated master key is persisted when the operator supplies no
+# ``ROUTSTR_SECRET_KEY``. Defaults beside the SQLite database so it rides whatever
+# volume already persists the data (a container recreate would otherwise generate
+# a fresh key and be unable to decrypt existing secrets); falls back to the
+# working directory when the DB location is unknown. Override the exact path with
+# ``ROUTSTR_SECRET_KEY_FILE``.
+_KEY_FILE_ENV = "ROUTSTR_SECRET_KEY_FILE"
+_DEFAULT_KEY_FILE = "routstr_secret.key"
+
+# Minimum admin-password length, enforced wherever a password is set/changed
+# (admin endpoints + the recovery script) so the policy lives in one place.
+MIN_PASSWORD_LENGTH = 8
+
+# scrypt parameters; packed into each hash so verification is parameter-free.
+_SCRYPT_N = 2**14
+_SCRYPT_R = 8
+_SCRYPT_P = 1
+_SCRYPT_DKLEN = 32
+_SCRYPT_SALT_BYTES = 16
+
+
+def _database_dir() -> Path | None:
+    """Directory of the SQLite database file, or ``None`` when it has no on-disk
+    location (a non-SQLite URL or ``:memory:``).
+
+    Read from ``DATABASE_URL`` at call time and parsed here rather than importing
+    ``routstr.core.db`` — that module builds the engine at import, which the
+    crypto layer must not drag in. Mirrors db.py's ``DATABASE_URL`` default.
+    """
+    url_str = os.environ.get("DATABASE_URL", "sqlite+aiosqlite:///keys.db")
+    try:
+        url = make_url(url_str)
+    except ArgumentError:
+        return None
+    if url.get_backend_name() != "sqlite" or not url.database:
+        return None
+    if url.database == ":memory:":
+        return None
+    return Path(url.database).parent
+
+
+def _key_file_path() -> Path:
+    """Where the auto-generated master key is read from / written to.
+
+    ``ROUTSTR_SECRET_KEY_FILE`` wins; otherwise the key sits beside the SQLite
+    database so it persists on the same volume as the data, falling back to the
+    working directory when the DB location is unknown.
+    """
+    override = os.environ.get(_KEY_FILE_ENV)
+    if override:
+        return Path(override)
+    directory = _database_dir()
+    return (directory or Path()) / _DEFAULT_KEY_FILE
+
+
+def _read_key_file(path: Path) -> str | None:
+    try:
+        stored = path.read_text().strip()
+    except OSError:
+        return None
+    if not stored:
+        return None
+    _repair_key_file_perms(path)
+    return stored
+
+
+def _repair_key_file_perms(path: Path) -> None:
+    # A master key must never be group/other-readable. On POSIX, tighten loose
+    # permissions to owner-only (0600) rather than trust — or hard-fail on — a
+    # world-readable key; a friendlier repair keeps an upgrading node booting.
+    if os.name != "posix":
+        return
+    try:
+        mode = path.stat().st_mode
+    except OSError:
+        return
+    if mode & 0o077:
+        try:
+            os.chmod(path, 0o600)
+        except OSError:
+            pass
+
+
+def _load_secret_key() -> str | None:
+    """The configured key without provisioning: env var, then the key file."""
+    return os.environ.get("ROUTSTR_SECRET_KEY") or _read_key_file(_key_file_path())
+
+
+def _warn_generated_key(path: Path) -> None:
+    # stdout, not the logger: the operator must see this once (e.g. in
+    # ``docker compose logs``), but it must never be persisted into the on-disk
+    # log files the logger also writes. Mirrors the generated-admin-password
+    # notice so an upgrade cannot silently create an unbacked key.
+    print(
+        "No ROUTSTR_SECRET_KEY was set; generated one to encrypt node secrets at "
+        f"rest and saved it to {path}.\n"
+        "!! BACK UP THIS FILE. If it is lost, the encrypted secrets cannot be "
+        "recovered and will have to be re-entered.\n"
+        "To manage the key yourself (e.g. from a secrets manager) set "
+        "ROUTSTR_SECRET_KEY in the environment instead; the value is in the file "
+        "above.",
+        flush=True,
+    )
+
+
+def _generate_and_persist_key(path: Path) -> str:
+    """Generate a Fernet key, persist it owner-only and atomically, warn once.
+
+    The key is written to a temp file in the same directory, flushed durably,
+    then ``os.link``-ed into place. ``os.link`` publishes the complete file in a
+    single atomic step — a crash mid-write leaves only the temp file (which is
+    removed), never a half-written or empty key at the final path that a later
+    boot would read as corrupt. It also refuses to overwrite an existing key, so
+    a racing worker that generated first keeps ownership (secrets may already be
+    encrypted under its key); the loser adopts that key instead of clobbering it.
+    """
+    key = Fernet.generate_key().decode()
+    path.parent.mkdir(parents=True, exist_ok=True)
+    fd, tmp_name = tempfile.mkstemp(
+        dir=path.parent, prefix=".routstr_secret.", suffix=".tmp"
+    )
+    tmp = Path(tmp_name)
+    try:
+        with os.fdopen(fd, "w") as handle:
+            handle.write(key)  # mkstemp already created it 0600
+            handle.flush()
+            os.fsync(handle.fileno())
+        try:
+            os.link(tmp, path)
+        except FileExistsError:
+            # A concurrent worker linked its key in first; adopt theirs rather
+            # than clobber a key that secrets may already be encrypted under.
+            existing = _read_key_file(path)
+            if existing:
+                return existing
+            raise
+        _fsync_dir(path.parent)
+    finally:
+        tmp.unlink(missing_ok=True)
+    _warn_generated_key(path)
+    return key
+
+
+def _fsync_dir(directory: Path) -> None:
+    # Persist the new directory entry so the linked key survives a crash right
+    # after publish. Best-effort: not every platform lets you fsync a directory.
+    try:
+        dir_fd = os.open(directory, os.O_RDONLY)
+    except OSError:
+        return
+    try:
+        os.fsync(dir_fd)
+    except OSError:
+        pass
+    finally:
+        os.close(dir_fd)
+
+
+def ensure_secret_key() -> str:
+    """Return the master key, provisioning one if the operator supplied none.
+
+    Precedence: the ``ROUTSTR_SECRET_KEY`` env var, then the persisted key file,
+    otherwise a freshly generated key written to the key file (with a one-time
+    operator notice). This keeps encryption at rest mandatory while letting an
+    existing node upgrade without setting a key first. A malformed env key is
+    left to fail at :func:`get_fernet` — it is an operator error, not an unset
+    key, so it must not trigger silent self-provisioning.
+    """
+    env_key = os.environ.get("ROUTSTR_SECRET_KEY")
+    if env_key:
+        return env_key
+    path = _key_file_path()
+    return _read_key_file(path) or _generate_and_persist_key(path)
+
+
+def _fernet_from_key(key: str) -> Fernet:
+    try:
+        return Fernet(key.encode())
+    except (ValueError, TypeError) as exc:
+        raise RuntimeError(
+            "ROUTSTR_SECRET_KEY is malformed; it must be a url-safe base64 "
+            "32-byte Fernet key. Generate one with:\n    " + _GEN_COMMAND
+        ) from exc
+
+
+def get_fernet() -> Fernet:
+    """Build a :class:`Fernet` from the configured key (env var or key file).
+
+    Strict: this never generates a key, so already-encrypted ciphertext is never
+    shadowed by a fresh key. A read with no key configured fails fast with the
+    generation command.
+    """
+    key = _load_secret_key()
+    if not key:
+        raise RuntimeError(
+            "ROUTSTR_SECRET_KEY is not set. It is required to encrypt secrets at "
+            "rest. Generate one with:\n    " + _GEN_COMMAND
+        )
+    return _fernet_from_key(key)
+
+
+def encrypt(plaintext: str) -> str:
+    """Encrypt ``plaintext`` into a self-describing ``fernet:v1:`` token.
+
+    Provisions a master key (env var, key file, or a freshly generated one) so an
+    upgrading node never has to set one before its first secret is stored; the
+    value is always encrypted, never persisted in plaintext.
+    """
+    fernet = _fernet_from_key(ensure_secret_key())
+    return _PREFIX + fernet.encrypt(plaintext.encode()).decode()
+
+
+def is_encrypted(value: str) -> bool:
+    """True if ``value`` carries the ``fernet:v1:`` prefix this module emits."""
+    return value.startswith(_PREFIX)
+
+
+def decrypt(ciphertext: str) -> str:
+    """Decrypt a ``fernet:v1:`` token.
+
+    Raises ``ValueError`` for an unprefixed value (so legacy plaintext is never
+    mistaken for ciphertext) and ``InvalidToken`` when the value was written
+    under a different ``ROUTSTR_SECRET_KEY``.
+    """
+    if not is_encrypted(ciphertext):
+        raise ValueError("value is not fernet:v1: ciphertext")
+    token = ciphertext[len(_PREFIX) :]
+    return get_fernet().decrypt(token.encode()).decode()
+
+
+def hash_password(password: str) -> str:
+    """Salted scrypt hash, self-describing as ``scrypt:n:r:p:salt:hash``."""
+    salt = secrets.token_bytes(_SCRYPT_SALT_BYTES)
+    derived = hashlib.scrypt(
+        password.encode(),
+        salt=salt,
+        n=_SCRYPT_N,
+        r=_SCRYPT_R,
+        p=_SCRYPT_P,
+        dklen=_SCRYPT_DKLEN,
+    )
+    return ":".join(
+        [
+            "scrypt",
+            str(_SCRYPT_N),
+            str(_SCRYPT_R),
+            str(_SCRYPT_P),
+            base64.b64encode(salt).decode(),
+            base64.b64encode(derived).decode(),
+        ]
+    )
+
+
+def verify_password(password: str, stored: str) -> bool:
+    """Constant-time check of ``password`` against a :func:`hash_password` value."""
+    try:
+        scheme, n, r, p, salt_b64, hash_b64 = stored.split(":")
+        if scheme != "scrypt":
+            return False
+        n_int, r_int, p_int = int(n), int(r), int(p)
+        # Cap the work factor at the parameters this module emits. scrypt's
+        # memory cost grows with N*r, so an oversized N/r in a tampered or
+        # corrupt stored hash could turn a single login into an OOM/DoS.
+        if n_int > _SCRYPT_N or r_int > _SCRYPT_R or p_int > _SCRYPT_P:
+            return False
+        salt = base64.b64decode(salt_b64)
+        expected = base64.b64decode(hash_b64)
+        derived = hashlib.scrypt(
+            password.encode(),
+            salt=salt,
+            n=n_int,
+            r=r_int,
+            p=p_int,
+            dklen=len(expected),
+        )
+    except (ValueError, TypeError):
+        return False
+    return hmac.compare_digest(derived, expected)

routstr/core/version.py

@@ -20,7 +20,7 @@ import subprocess
 from functools import lru_cache
 from pathlib import Path
 
-BASE_VERSION = "0.4.3"
+BASE_VERSION = "0.4.4"
 
 _REPO_ROOT = Path(__file__).resolve().parents[2]
 _GIT_TIMEOUT_SECONDS = 2.0

routstr/lightning.py

@@ -5,13 +5,14 @@ import time
 
 from fastapi import APIRouter, Depends, Header, HTTPException
 from pydantic import BaseModel, Field
-from sqlmodel import col, select
+from sqlalchemy.orm.attributes import set_committed_value
+from sqlmodel import col, select, update
 from sqlmodel.ext.asyncio.session import AsyncSession
 
 from .core.db import ApiKey, LightningInvoice, create_session, get_session
 from .core.logging import get_logger
 from .core.settings import settings
-from .wallet import get_wallet
+from .wallet import get_wallet, wallet_operation_guard
 
 logger = get_logger(__name__)
 
@@ -222,80 +223,180 @@ async def recover_invoice(
 async def check_invoice_payment(
     invoice: LightningInvoice, session: AsyncSession
 ) -> None:
+    # Minting makes proofs visible before database finalization. Share the
+    # cross-process wallet guard with owner payout so that visibility and the
+    # corresponding liability commit are observed atomically by the payout loop.
+    async with wallet_operation_guard():
+        await _check_invoice_payment_locked(invoice, session)
+
+
+async def _check_invoice_payment_locked(
+    invoice: LightningInvoice, session: AsyncSession
+) -> None:
+    minted = False
+    invoice_id = invoice.id
+    invoice_purpose = invoice.purpose
+    invoice_amount_sats = invoice.amount_sats
+    invoice_payment_hash = invoice.payment_hash
+    finalized_api_key_hash = invoice.api_key_hash
     try:
+        # A preceding invoice lookup starts a transaction. End it before the
+        # potentially slow mint request so it cannot pin a pool connection.
+        await session.commit()
+
         wallet = await get_wallet(settings.primary_mint, "sat")
+        mint_status = await wallet.get_mint_quote(invoice_payment_hash)
+        if not mint_status.paid:
+            return
 
-        mint_status = await wallet.get_mint_quote(invoice.payment_hash)
+        # Do not redeem a paid top-up quote if its target has already been
+        # pruned. This validation owns a short-lived session and releases its
+        # connection before mint redemption starts.
+        if invoice_purpose == "topup":
+            if not finalized_api_key_hash:
+                raise ValueError("No API key associated with topup invoice")
+            async with create_session() as validation_session:
+                target = await validation_session.get(ApiKey, finalized_api_key_hash)
+                if target is None:
+                    terminal = await validation_session.exec(  # type: ignore[call-overload]
+                        update(LightningInvoice)
+                        .where(
+                            col(LightningInvoice.id) == invoice_id,
+                            col(LightningInvoice.status) == "pending",
+                        )
+                        .values(status="reconciliation_required")
+                    )
+                    await validation_session.commit()
+                    if terminal.rowcount == 1:
+                        set_committed_value(
+                            invoice, "status", "reconciliation_required"
+                        )
+                    else:
+                        committed_invoice = await validation_session.get(
+                            LightningInvoice, invoice_id
+                        )
+                        if committed_invoice is not None:
+                            set_committed_value(
+                                invoice, "status", committed_invoice.status
+                            )
+                    logger.critical(
+                        "Paid topup invoice target API key was not found; reconciliation required",
+                        extra={"invoice_id": invoice_id},
+                    )
+                    return
 
-        if mint_status.paid:
-            invoice.status = "paid"
-            invoice.paid_at = int(time.time())
+        # The mint enforces single-use quotes, so a concurrent checker that
+        # races us here fails inside wallet.mint rather than double-minting.
+        await wallet.mint(invoice_amount_sats, quote_id=invoice_payment_hash)
+        minted = True
 
-            if invoice.purpose == "create":
-                api_key = await create_api_key_from_invoice(invoice, session)
-                invoice.api_key_hash = api_key.hashed_key
-            elif invoice.purpose == "topup" and invoice.api_key_hash:
-                await topup_api_key_from_invoice(invoice, session)
+        # Paid finalization owns a fresh session. The API/watcher session is
+        # never rolled back by this function, so its invoice and sibling ORM
+        # objects remain usable after a DB failure or lost CAS race.
+        async with create_session() as finalization_session:
+            if invoice_purpose == "create":
+                api_key = await _create_api_key_record(invoice, finalization_session)
+                finalized_api_key_hash = api_key.hashed_key
+            elif invoice_purpose == "topup":
+                await _credit_topup_record(invoice, finalization_session)
 
-            await session.commit()
-
-            logger.info(
-                "Lightning invoice paid",
-                extra={
-                    "invoice_id": invoice.id,
-                    "amount_sats": invoice.amount_sats,
-                    "purpose": invoice.purpose,
-                    "api_key_hash": invoice.api_key_hash[:8] + "..."
-                    if invoice.api_key_hash
-                    else None,
-                },
+            # Conditional transition guards against double-credit: the credit
+            # above and this status flip commit atomically, and a lost race
+            # rolls both back in the owned finalization session.
+            paid_at = int(time.time())
+            finalized = await finalization_session.exec(  # type: ignore[call-overload]
+                update(LightningInvoice)
+                .where(
+                    col(LightningInvoice.id) == invoice_id,
+                    col(LightningInvoice.status) == "pending",
+                )
+                .values(
+                    status="paid",
+                    paid_at=paid_at,
+                    api_key_hash=finalized_api_key_hash,
+                )
             )
+            if finalized.rowcount != 1:
+                await finalization_session.rollback()
+                committed_invoice = await finalization_session.get(
+                    LightningInvoice, invoice_id
+                )
+                await finalization_session.commit()
+                if committed_invoice is not None:
+                    # A concurrent finalizer won the CAS. Publish only the
+                    # state observed from the database after ending the owned
+                    # read transaction; never refresh the caller's session.
+                    set_committed_value(
+                        invoice, "api_key_hash", committed_invoice.api_key_hash
+                    )
+                    set_committed_value(invoice, "status", committed_invoice.status)
+                    set_committed_value(invoice, "paid_at", committed_invoice.paid_at)
+                return
+            await finalization_session.commit()
 
-    except Exception as e:
+        # Only publish finalized values to the caller-owned object after the
+        # owned transaction has committed successfully.
+        set_committed_value(invoice, "api_key_hash", finalized_api_key_hash)
+        set_committed_value(invoice, "status", "paid")
+        set_committed_value(invoice, "paid_at", paid_at)
+
+        logger.info(
+            "Lightning invoice paid",
+            extra={
+                "invoice_id": invoice_id,
+                "amount_sats": invoice_amount_sats,
+                "purpose": invoice_purpose,
+                "api_key_hash": finalized_api_key_hash[:8] + "..."
+                if finalized_api_key_hash
+                else None,
+            },
+        )
+    except BaseException as e:
+        # BaseException so task cancellation (e.g. client disconnect) after a
+        # successful mint still triggers the reconciliation alert. Any rollback
+        # belongs to create_session(), never to the caller-owned session.
+        if minted:
+            logger.critical(
+                "Invoice mint succeeded but DB finalization failed; reconciliation required",
+                extra={"invoice_id": invoice_id, "purpose": invoice_purpose},
+            )
+        if not isinstance(e, Exception):
+            raise
         logger.error(f"Failed to check invoice payment: {e}")
 
 
-async def create_api_key_from_invoice(
+async def _create_api_key_record(
     invoice: LightningInvoice, session: AsyncSession
 ) -> ApiKey:
-    wallet = await get_wallet(settings.primary_mint, "sat")
-    await wallet.mint(invoice.amount_sats, quote_id=invoice.payment_hash)
-
     dummy_token = f"invoice-{invoice.id}-{invoice.payment_hash}"
     hashed_key = hashlib.sha256(dummy_token.encode()).hexdigest()
-
     api_key = ApiKey(
         hashed_key=hashed_key,
-        balance=invoice.amount_sats * 1000,  # Convert to msats
+        balance=invoice.amount_sats * 1000,
         refund_currency="sat",
         refund_mint_url=settings.primary_mint,
         balance_limit=invoice.balance_limit,
         balance_limit_reset=invoice.balance_limit_reset,
         validity_date=invoice.validity_date,
     )
-
     session.add(api_key)
     await session.flush()
-
     return api_key
 
 
-async def topup_api_key_from_invoice(
+async def _credit_topup_record(
     invoice: LightningInvoice, session: AsyncSession
 ) -> None:
-    wallet = await get_wallet(settings.primary_mint, "sat")
-    await wallet.mint(invoice.amount_sats, quote_id=invoice.payment_hash)
-
     if not invoice.api_key_hash:
         raise ValueError("No API key associated with topup invoice")
-
-    api_key = await session.get(ApiKey, invoice.api_key_hash)
-    if not api_key:
+    credited = await session.exec(  # type: ignore[call-overload]
+        update(ApiKey)
+        .where(col(ApiKey.hashed_key) == invoice.api_key_hash)
+        .values(balance=col(ApiKey.balance) + invoice.amount_sats * 1000)
+    )
+    if credited.rowcount != 1:
         raise ValueError("Associated API key not found")
 
-    api_key.balance += invoice.amount_sats * 1000  # Convert to msats
-    await session.flush()
-
 
 INVOICE_WATCH_INTERVAL_SECONDS = 5
 INVOICE_WATCH_BATCH_LIMIT = 100

routstr/payment/cost_calculation.py

@@ -1,11 +1,23 @@
 import math
+from typing import TYPE_CHECKING
 
 from pydantic.v1 import BaseModel
 
 from ..core import get_logger
-from ..core.db import AsyncSession
 from ..core.settings import settings
 from .price import sats_usd_price
+from .usage import normalize_usage, parse_token_count
+
+if TYPE_CHECKING:
+    from .models import Model
+
+__all__ = [
+    "CostData",
+    "CostDataError",
+    "MaxCostData",
+    "calculate_cost",
+    "parse_token_count",
+]
 
 logger = get_logger(__name__)
 
@@ -33,17 +45,54 @@ class CostDataError(BaseModel):
     code: str
 
 
+def _empty_cost(cls: type[CostData] = CostData) -> CostData:
+    """Build an all-zero cost object — a full refund for an empty response.
+
+    Shared by the two paths that must not bill: an upstream response with no
+    usage data at all, and one that reports a USD cost but carries zero tokens
+    in every bucket.
+    """
+    return cls(
+        base_msats=0,
+        input_msats=0,
+        output_msats=0,
+        total_msats=0,
+        total_usd=0.0,
+        input_tokens=0,
+        output_tokens=0,
+        cache_read_input_tokens=0,
+        cache_creation_input_tokens=0,
+        cache_read_msats=0,
+        cache_creation_msats=0,
+    )
+
+
 async def calculate_cost(
-    response_data: dict, max_cost: int, session: AsyncSession
+    response_data: dict,
+    max_cost: int,
+    model_obj: "Model | None" = None,
+    provider_fee: float | None = None,
 ) -> CostData | MaxCostData | CostDataError:
     """Calculate the cost of an API request based on token usage.
 
     Args:
         response_data: Response data containing usage information
         max_cost: Maximum cost in millisats
+        model_obj: The model that actually served the request. When given,
+            its pricing is billed directly; without it, pricing is re-derived
+            from the response's model string via the alias map, which resolves
+            to the best-ranked candidate — not necessarily the serving one.
+        provider_fee: The serving provider's fee multiplier, applied on the
+            USD-cost path and the litellm pricing fallback (configured model
+            pricing already carries the fee baked in). Without it, the fee is
+            re-derived from the response's model string, which yields the
+            best-ranked provider's fee.
 
     Returns:
         Cost data or error information
+
+    The response's usage object is normalized with the default union parser;
+    this function holds no vendor-dialect knowledge of its own.
     """
     logger.debug(
         "Starting cost calculation",
@@ -54,8 +103,9 @@ async def calculate_cost(
         },
     )
 
-    # Check for usage data
-    if "usage" not in response_data or response_data["usage"] is None:
+    usage = normalize_usage(response_data.get("usage"))
+
+    if usage is None:
         logger.warning(
             "No usage data in response — billing at MaxCostData with zero "
             "tokens. Dashboard will show this request as `(0+0)`. Most "
@@ -70,34 +120,41 @@ async def calculate_cost(
                 else None,
             },
         )
-        return MaxCostData(
-            base_msats=0,
-            input_msats=0,
-            output_msats=0,
-            total_msats=0,
-            total_usd=0.0,
-            input_tokens=0,
-            output_tokens=0,
-            cache_read_input_tokens=0,
-            cache_creation_input_tokens=0,
-            cache_read_msats=0,
-            cache_creation_msats=0,
-        )
+        return _empty_cost(MaxCostData)
 
-    usage_data = response_data["usage"]
+    usage_data = response_data.get("usage") or {}
+    if not isinstance(usage_data, dict):
+        usage_data = {}
 
-    # Extract token counts
-    input_tokens = _extract_token_pair(usage_data, "prompt_tokens", "input_tokens")
-    output_tokens = _extract_token_pair(usage_data, "completion_tokens", "output_tokens")
-
-    # Extract cache tokens (handles OpenAI vs Anthropic formats)
-    cache_read_tokens, cache_creation_tokens, input_tokens = _extract_cache_tokens(
-        usage_data, input_tokens
-    )
+    input_tokens = usage.input_tokens
+    output_tokens = usage.output_tokens
+    cache_read_tokens = usage.cache_read_tokens
+    cache_creation_tokens = usage.cache_write_tokens
 
     # Try USD cost first
     usd_cost = _resolve_usd_cost(usage_data, response_data)
     if usd_cost > 0:
+        truly_empty = (
+            input_tokens == 0
+            and output_tokens == 0
+            and cache_read_tokens == 0
+            and cache_creation_tokens == 0
+        )
+        if truly_empty:
+            logger.warning(
+                "Upstream reported a USD cost but the response carries no "
+                "tokens at all (input, output, cache-read and cache-creation "
+                "are all zero) — refunding in full rather than billing the "
+                "USD-derived cost for an empty response.",
+                extra={
+                    "model": response_data.get("model", "unknown"),
+                    "usd_cost": usd_cost,
+                    "usage_keys": sorted(usage_data.keys())
+                    if isinstance(usage_data, dict)
+                    else None,
+                },
+            )
+            return _empty_cost()
         if input_tokens == 0 and output_tokens == 0:
             logger.warning(
                 "Upstream reported a USD cost but no token counts — "
@@ -115,11 +172,16 @@ async def calculate_cost(
                 },
             )
         try:
+            cost_details = usage_data.get("cost_details", {})
+            if not isinstance(cost_details, dict):
+                cost_details = {}
             input_usd = _coerce_usd(
-                usage_data.get("cost_details", {}).get("input_cost", 0)
+                cost_details.get("input_cost")
+                or cost_details.get("upstream_inference_prompt_cost")
             )
             output_usd = _coerce_usd(
-                usage_data.get("cost_details", {}).get("output_cost", 0)
+                cost_details.get("output_cost")
+                or cost_details.get("upstream_inference_completions_cost")
             )
             return _calculate_from_usd_cost(
                 usd_cost,
@@ -130,6 +192,7 @@ async def calculate_cost(
                 cache_creation_tokens,
                 output_tokens,
                 response_data,
+                provider_fee,
             )
         except Exception as e:
             logger.warning(
@@ -143,7 +206,7 @@ async def calculate_cost(
 
     # Fall back to token-based pricing
     try:
-        pricing_rates = _get_pricing_rates(response_data)
+        pricing_rates = _get_pricing_rates(response_data, model_obj, provider_fee)
     except ValueError as e:
         return CostDataError(message=str(e), code="pricing_error")
 
@@ -202,22 +265,6 @@ async def calculate_cost(
 # ============================================================================
 
 
-def parse_token_count(value: object) -> int:
-    """Parse a token count from various formats (int, float, str, bool)."""
-    if isinstance(value, bool):
-        return 0
-    if isinstance(value, int):
-        return max(0, value)
-    if isinstance(value, float):
-        return max(0, int(value))
-    if isinstance(value, str):
-        try:
-            return max(0, int(float(value)))
-        except ValueError:
-            return 0
-    return 0
-
-
 def _coerce_usd(value: object) -> float:
     """Coerce a value to USD float, handling various formats safely."""
     if value is None or isinstance(value, bool):
@@ -230,41 +277,20 @@ def _coerce_usd(value: object) -> float:
         return 0.0
 
 
-def _extract_token_pair(
-    usage_data: dict, standard_field: str, alt_field: str
-) -> int:
-    """Extract token count trying two field names in order."""
-    value = parse_token_count(usage_data.get(standard_field, 0))
-    if value > 0:
-        return value
-    return parse_token_count(usage_data.get(alt_field, 0))
-
-
-def _extract_cache_tokens(usage_data: dict, input_tokens: int) -> tuple[int, int, int]:
-    """Extract cache tokens, handling OpenAI vs Anthropic formats.
-
-    Returns: (cache_read_tokens, cache_creation_tokens, adjusted_input_tokens)
-    """
-    cache_read = parse_token_count(usage_data.get("cache_read_input_tokens", 0))
-    cache_creation = parse_token_count(
-        usage_data.get("cache_creation_input_tokens", 0)
-    )
-
-    # OpenAI: cache is included in input_tokens, subtract it
-    prompt_details = usage_data.get("prompt_tokens_details")
-    if isinstance(prompt_details, dict) and not cache_read:
-        openai_cached = parse_token_count(prompt_details.get("cached_tokens", 0))
-        if openai_cached:
-            cache_read = openai_cached
-            input_tokens = max(0, input_tokens - cache_read)
-
-    return cache_read, cache_creation, input_tokens
-
-
 def _resolve_usd_cost(usage_data: dict, response_data: dict) -> float:
     """Resolve USD cost with clear priority order.
 
-    Priority: cost_details.total_cost → total_cost → cost (in both usage and response).
+    Priority:
+
+    1. ``cost_details.total_cost``
+    2. ``cost_details.upstream_inference_cost`` (BYOK — see below)
+    3. ``total_cost`` → ``cost`` (in both usage and response)
+
+    **BYOK path (PPQ.AI):** when ``is_byok`` is true the ``usage.cost`` field
+    is only a small (~5 %) routing fee, not the inference cost.  The real cost
+    lives in ``cost_details.upstream_inference_cost`` and the provider's
+    balance is debited by ``upstream_inference_cost + byok_fee``.  Billing just
+    the fee under-charges by ~20×.
     """
     cost_details = usage_data.get("cost_details")
     if isinstance(cost_details, dict):
@@ -272,6 +298,18 @@ def _resolve_usd_cost(usage_data: dict, response_data: dict) -> float:
         if cost > 0:
             return cost
 
+        # PPQ.AI BYOK: upstream_inference_cost is the real inference cost;
+        # usage.cost is only a ~5 % BYOK routing fee.  Bill the sum — what PPQ
+        # actually deducts from the balance.  For non-BYOK providers (e.g.
+        # OpenRouter) usage.cost already equals upstream_inference_cost, so we
+        # fall through to the normal ``cost`` lookup below.
+        upstream_cost = _coerce_usd(
+            cost_details.get("upstream_inference_cost")
+        )
+        if upstream_cost > 0 and usage_data.get("is_byok"):
+            byok_fee = _coerce_usd(usage_data.get("cost"))
+            return upstream_cost + byok_fee
+
     for source in [usage_data, response_data]:
         if not isinstance(source, dict):
             continue
@@ -285,55 +323,106 @@ def _resolve_usd_cost(usage_data: dict, response_data: dict) -> float:
 
 def _get_pricing_rates(
     response_data: dict,
+    model_obj: "Model | None",
+    provider_fee: float | None,
 ) -> tuple[float, float, float, float] | None:
-    """Get model-based pricing rates or None if using fixed pricing.
+    """Get configured rates, falling back to LiteLLM's model cost map.
 
-    Returns: (input_rate, output_rate, cache_read_rate, cache_write_rate)
+    The served ``model_obj`` (when the caller has it) is billed directly;
+    otherwise the response's model string is resolved through the alias map,
+    which yields the best-ranked candidate rather than the serving one.
+
+    Returns: (input_rate, output_rate, cache_read_rate, cache_write_rate).
+    ``None`` means configured fixed pricing should be used by the caller.
     """
-    if settings.fixed_pricing:
+    if settings.fixed_pricing and (
+        settings.fixed_per_1k_input_tokens
+        or settings.fixed_per_1k_output_tokens
+    ):
         return None
 
     from ..proxy import get_model_instance
+    from .models import litellm_cost_entry
 
     response_model = response_data.get("model", "")
-    model_obj = get_model_instance(response_model)
-
-    if not model_obj:
-        logger.error("Invalid model in response", extra={"response_model": response_model})
-        raise ValueError(f"Invalid model: {response_model}")
-
-    if not model_obj.sats_pricing:
-        logger.error(
-            "Model pricing not defined",
-            extra={"model": response_model, "model_id": response_model},
+    if model_obj is None:
+        logger.warning(
+            "Settling without routed model identity — re-deriving pricing "
+            "from the response's model string via the alias map",
+            extra={"response_model": response_model},
         )
-        raise ValueError("Model pricing not defined")
+        model_obj = get_model_instance(response_model)
 
-    try:
-        mspp = float(model_obj.sats_pricing.prompt)
-        mspc = float(model_obj.sats_pricing.completion)
-        mscr = float(model_obj.sats_pricing.input_cache_read or 0)
-        mscw = float(model_obj.sats_pricing.input_cache_write or 0)
+    if model_obj and model_obj.sats_pricing:
+        try:
+            mspp = float(model_obj.sats_pricing.prompt)
+            mspc = float(model_obj.sats_pricing.completion)
+            mscr = float(model_obj.sats_pricing.input_cache_read or 0)
+            mscw = float(model_obj.sats_pricing.input_cache_write or 0)
 
-        mspp_1k = mspp * 1_000_000.0
-        mspc_1k = mspc * 1_000_000.0
-        mscr_1k = mscr * 1_000_000.0 if mscr > 0 else mspp_1k
-        mscw_1k = mscw * 1_000_000.0 if mscw > 0 else mspp_1k
+            mspp_1k = mspp * 1_000_000.0
+            mspc_1k = mspc * 1_000_000.0
+            mscr_1k = mscr * 1_000_000.0 if mscr > 0 else mspp_1k
+            mscw_1k = mscw * 1_000_000.0 if mscw > 0 else mspp_1k
+            source = "configured"
+        except Exception as e:
+            logger.error("Invalid pricing data", extra={"error": str(e)})
+            raise ValueError("Invalid pricing data") from e
+    else:
+        pricing_model = (
+            model_obj.forwarded_model_id if model_obj else None
+        ) or response_model
+        pricing = litellm_cost_entry(pricing_model)
+        if pricing is None:
+            logger.error(
+                "Model pricing not found in configured models or LiteLLM",
+                extra={
+                    "response_model": response_model,
+                    "pricing_model": pricing_model,
+                },
+            )
+            raise ValueError(f"Pricing not found for model: {response_model}")
 
-        logger.info(
-            "Applied model-specific pricing",
-            extra={
-                "model": response_model,
-                "input_price_msats_per_1k": mspp_1k,
-                "output_price_msats_per_1k": mspc_1k,
-                "cache_read_price_msats_per_1k": mscr_1k,
-                "cache_write_price_msats_per_1k": mscw_1k,
-            },
+        input_usd = _coerce_usd(pricing.get("input_cost_per_token"))
+        output_usd = _coerce_usd(pricing.get("output_cost_per_token"))
+        if input_usd <= 0 or output_usd <= 0:
+            raise ValueError(f"Incomplete LiteLLM pricing for model: {pricing_model}")
+
+        if provider_fee is None:
+            provider_fee = _resolve_provider_fee(response_model)
+        usd_per_sat = sats_usd_price()
+        mspp_1k = input_usd * provider_fee * 1_000_000.0 / usd_per_sat
+        mspc_1k = output_usd * provider_fee * 1_000_000.0 / usd_per_sat
+        cache_read_usd = _coerce_usd(
+            pricing.get("cache_read_input_token_cost")
         )
-        return mspp_1k, mspc_1k, mscr_1k, mscw_1k
-    except Exception as e:
-        logger.error("Invalid pricing data", extra={"error": str(e)})
-        raise ValueError("Invalid pricing data") from e
+        cache_write_usd = _coerce_usd(
+            pricing.get("cache_creation_input_token_cost")
+        )
+        mscr_1k = (
+            cache_read_usd * provider_fee * 1_000_000.0 / usd_per_sat
+            if cache_read_usd > 0
+            else mspp_1k
+        )
+        mscw_1k = (
+            cache_write_usd * provider_fee * 1_000_000.0 / usd_per_sat
+            if cache_write_usd > 0
+            else mspp_1k
+        )
+        source = "litellm"
+
+    logger.info(
+        "Applied model-specific pricing",
+        extra={
+            "model": response_model,
+            "pricing_source": source,
+            "input_price_msats_per_1k": mspp_1k,
+            "output_price_msats_per_1k": mspc_1k,
+            "cache_read_price_msats_per_1k": mscr_1k,
+            "cache_write_price_msats_per_1k": mscw_1k,
+        },
+    )
+    return mspp_1k, mspc_1k, mscr_1k, mscw_1k
 
 
 def _resolve_provider_fee(model_id: str) -> float:
@@ -361,9 +450,11 @@ def _calculate_from_usd_cost(
     cache_creation_tokens: int,
     output_tokens: int,
     response_data: dict,
+    provider_fee: float | None,
 ) -> CostData:
     """Calculate cost from USD figures, deriving input/output split from tokens."""
-    provider_fee = _resolve_provider_fee(response_data.get("model", ""))
+    if provider_fee is None:
+        provider_fee = _resolve_provider_fee(response_data.get("model", ""))
     usd_cost = usd_cost * provider_fee
     input_usd = input_usd * provider_fee
     output_usd = output_usd * provider_fee
@@ -372,8 +463,12 @@ def _calculate_from_usd_cost(
     cost_in_msats = math.ceil(cost_in_sats * 1000)
 
     if input_usd > 0 or output_usd > 0:
-        input_msats = int((input_usd * sats_per_usd) * 1000)
-        output_msats = int((output_usd * sats_per_usd) * 1000)
+        # The total is the authoritative billed amount. Allocating that integer
+        # total proportionally avoids losing sub-millisatoshi remainders when
+        # input and output components are each truncated independently.
+        component_usd = input_usd + output_usd
+        input_msats = math.floor(cost_in_msats * input_usd / component_usd)
+        output_msats = cost_in_msats - input_msats
     else:
         effective_input_tokens = (
             input_tokens + cache_read_tokens + cache_creation_tokens
@@ -454,10 +549,21 @@ def _calculate_from_tokens(
         },
     )
 
+    # Fold the cache-read/write cost into the visible ``input_msats`` so a
+    # dashboard that renders I / O / T sees ``input + output == total``
+    # exactly. This mirrors ``_fold_cache_into_input_tokens`` (which rolls the
+    # cache token counts into the visible prompt total). The standalone
+    # ``cache_read_msats`` / ``cache_creation_msats`` fields stay populated for
+    # clients that want the breakdown; nothing sums the components to derive
+    # ``total_msats`` (it is computed independently above), so this is
+    # display-only and does not change what is billed.
+    visible_output_msats = int(calc_output_msats)
+    visible_input_msats = token_based_cost - visible_output_msats
+
     return CostData(
         base_msats=0,
-        input_msats=int(calc_input_msats),
-        output_msats=int(calc_output_msats),
+        input_msats=visible_input_msats,
+        output_msats=visible_output_msats,
         total_msats=token_based_cost,
         total_usd=total_usd,
         input_tokens=input_tokens,

routstr/payment/helpers.py

@@ -11,6 +11,8 @@ from PIL import Image
 from sqlmodel.ext.asyncio.session import AsyncSession
 
 from ..core import get_logger
+from ..core.exceptions import UpstreamError
+from ..core.redaction import redact_org_ids
 from ..core.settings import settings
 from ..wallet import deserialize_token_from_string
 
@@ -418,16 +420,27 @@ def create_error_response(
     status_code: int,
     request: Request,
     token: str | None = None,
+    code: str | int | None = None,
+    details: dict[str, object] | None = None,
 ) -> Response:
-    """Create a standardized error response."""
+    """Create a standardized error response.
+
+    ``code`` is a stable, machine-readable classification (e.g.
+    ``UPSTREAM_RATE_LIMIT``); when omitted it defaults to the HTTP status code
+    for backwards compatibility. ``details`` carries optional structured,
+    redaction-safe context.
+    """
+    error_obj: dict[str, object] = {
+        "message": redact_org_ids(message),
+        "type": error_type,
+        "code": code if code is not None else status_code,
+    }
+    if details is not None:
+        error_obj["details"] = details
     return Response(
         content=json.dumps(
             {
-                "error": {
-                    "message": message,
-                    "type": error_type,
-                    "code": status_code,
-                },
+                "error": error_obj,
                 "request_id": getattr(request.state, "request_id", "unknown"),
             }
         ),
@@ -435,3 +448,20 @@ def create_error_response(
         media_type="application/json",
         headers={"X-Cashu": token} if token else {},
     )
+
+
+def create_upstream_error_response(
+    error: UpstreamError,
+    request: Request,
+    fallback_status: int = 502,
+) -> Response:
+    """Build an error response from an :class:`UpstreamError`, preserving its
+    structured ``code``, ``details``, and original ``status_code``."""
+    return create_error_response(
+        "upstream_error",
+        str(error),
+        error.status_code or fallback_status,
+        request=request,
+        code=getattr(error, "code", None),
+        details=getattr(error, "details", None),
+    )

routstr/payment/lnurl.py

@@ -1,11 +1,17 @@
 from __future__ import annotations
 
+import asyncio
 import math
 from typing import TypedDict
 
 import httpx
 from cashu.wallet.wallet import Proof, Wallet
 
+# The Cashu library issues POST /v1/melt/bolt11 with timeout=None, so a hung or
+# very slow mint can block a melt (and any caller, e.g. the payout loop)
+# indefinitely. Bound it here so callers fail instead of hanging forever.
+MELT_TIMEOUT_SECONDS = 60
+
 try:
     from bech32 import bech32_decode, convertbits  # type: ignore
 except ModuleNotFoundError:  # pragma: no cover – allow runtime miss
@@ -220,10 +226,18 @@ async def raw_send_to_lnurl(
     if amount:
         proofs, _ = await wallet.select_to_send(proofs, amount, set_reserved=True)
 
-    _ = await wallet.melt(
-        proofs=proofs,
-        invoice=bolt11_invoice,
-        fee_reserve_sat=melt_quote_resp.fee_reserve,
-        quote_id=melt_quote_resp.quote,
-    )
+    try:
+        _ = await asyncio.wait_for(
+            wallet.melt(
+                proofs=proofs,
+                invoice=bolt11_invoice,
+                fee_reserve_sat=melt_quote_resp.fee_reserve,
+                quote_id=melt_quote_resp.quote,
+            ),
+            timeout=MELT_TIMEOUT_SECONDS,
+        )
+    except asyncio.TimeoutError as e:
+        raise LNURLError(
+            f"Melt timed out after {MELT_TIMEOUT_SECONDS}s (mint unresponsive)"
+        ) from e
     return final_amount

routstr/payment/models.py

@@ -85,6 +85,64 @@ class Model(BaseModel):
         return hash(self.id)
 
 
+def litellm_cost_entry(model_id: str) -> dict | None:
+    """Look up ``model_id`` in litellm's bundled cost map.
+
+    litellm ships per-model USD rates keyed by the exact OpenRouter id
+    (``deepseek/deepseek-chat``) or the bare model name (``gpt-4o``,
+    ``claude-sonnet-4-5``), so both spellings are tried. Keys are lowercase, so
+    a mixed-case upstream id (``deepseek-ai/DeepSeek-V4-Flash``) is retried via
+    a case-insensitive scan. Returns the matched cost dict, or ``None``.
+    """
+    import litellm
+
+    candidates = (model_id, model_id.split("/", 1)[-1])
+    for key in candidates:
+        info = litellm.model_cost.get(key)
+        if isinstance(info, dict):
+            return info
+
+    lowered = {c.lower() for c in candidates}
+    for key, info in litellm.model_cost.items():
+        if isinstance(key, str) and key.lower() in lowered and isinstance(info, dict):
+            return info
+    return None
+
+
+def backfill_cache_pricing(model_id: str, pricing: Pricing) -> Pricing:
+    """Fill missing cache rates from litellm's bundled cost map.
+
+    The OpenRouter model feed omits ``input_cache_read``/``input_cache_write``
+    for many models (most DeepSeek entries, openai/gpt-4o, ...). Without a
+    cache rate, billing falls back to the full input rate, which overcharges
+    cache reads (DeepSeek hits are 10x cheaper) and undercharges Anthropic
+    cache writes (1.25x). The lookup (see ``litellm_cost_entry``) tries both
+    id spellings and a case-insensitive fallback.
+
+    Rates already present (e.g. provided by OpenRouter) are authoritative and
+    never overwritten. Unknown models are returned unchanged.
+    """
+    needs_read = (pricing.input_cache_read or 0.0) <= 0.0
+    needs_write = (pricing.input_cache_write or 0.0) <= 0.0
+    if not (needs_read or needs_write):
+        return pricing
+
+    info = litellm_cost_entry(model_id)
+    if info is None:
+        return pricing
+
+    updated = Pricing.parse_obj(pricing.dict())
+    if needs_read:
+        read_rate = info.get("cache_read_input_token_cost")
+        if isinstance(read_rate, (int, float)) and read_rate > 0:
+            updated.input_cache_read = float(read_rate)
+    if needs_write:
+        write_rate = info.get("cache_creation_input_token_cost")
+        if isinstance(write_rate, (int, float)) and write_rate > 0:
+            updated.input_cache_write = float(write_rate)
+    return updated
+
+
 def _has_valid_pricing(model: dict) -> bool:
     """Check if model has valid pricing (not free, no negative values)."""
     pricing = model.get("pricing", {})
@@ -173,13 +231,29 @@ def _row_to_model(
     )
     top_provider_dict = json.loads(row.top_provider) if row.top_provider else None
 
-    if apply_provider_fee and isinstance(pricing, dict):
-        pricing = {k: float(v) * provider_fee for k, v in pricing.items()}
-
     if isinstance(pricing, dict) and float(pricing.get("request", 0.0)) <= 0.0:
         pricing["request"] = max(pricing.get("request", 0.0), 0.0)
 
     parsed_pricing = Pricing.parse_obj(pricing)
+
+    # Fill missing cache-read/write rates from litellm's cost map BEFORE applying
+    # the provider fee, so they carry the same markup as every other component.
+    # DB-stored override pricing (e.g. generic providers) omits cache rates;
+    # without this, ``_row_to_model`` bills cache reads at the full input rate —
+    # the ``_apply_provider_fee_to_model`` path backfills, but the override path
+    # used for admin-configured providers did not.
+    #
+    # Key on ``forwarded_model_id`` (the actual upstream model name litellm
+    # prices) when set: an alias row (id="local-alias",
+    # forwarded_model_id="deepseek-v4-flash") would otherwise look up the alias
+    # and miss the cache rate.
+    pricing_model_id = getattr(row, "forwarded_model_id", None) or row.id
+    parsed_pricing = backfill_cache_pricing(pricing_model_id, parsed_pricing)
+
+    if apply_provider_fee:
+        parsed_pricing = Pricing.parse_obj(
+            {k: float(v) * provider_fee for k, v in parsed_pricing.dict().items()}
+        )
     model = Model(
         id=row.id,
         name=row.name,

routstr/payment/usage.py

@@ -0,0 +1,131 @@
+"""Vendor-agnostic normalization of upstream usage objects.
+
+Upstream providers report token usage in vendor dialects that differ in field
+names and in whether cached tokens are included in the input count:
+
+* OpenAI / Azure / xAI / Groq / Moonshot / Qwen / Gemini-compat: cache reads in
+  ``prompt_tokens_details.cached_tokens``, included in ``prompt_tokens``.
+* OpenRouter: same as OpenAI plus cache *writes* in
+  ``prompt_tokens_details.cache_write_tokens``, also included in
+  ``prompt_tokens``.
+* litellm-normalized: same nesting, but names the write field
+  ``prompt_tokens_details.cache_creation_tokens`` (and additionally mirrors the
+  Anthropic top-level fields), with ``prompt_tokens`` as the grand total.
+* Anthropic native: ``cache_read_input_tokens`` / ``cache_creation_input_tokens``
+  top-level, additive to (not included in) ``input_tokens``.
+* DeepSeek: ``prompt_cache_hit_tokens`` / ``prompt_cache_miss_tokens``, with
+  ``prompt_tokens = hit + miss``.
+
+What decides whether cached tokens must be subtracted out of the input count is
+**which prompt field the vendor uses**, not which cache field appears:
+
+* ``prompt_tokens`` present -> cached + cache-write tokens are *included* in it
+  (OpenAI family, DeepSeek, OpenRouter, litellm); subtract both so
+  ``input_tokens`` holds only the regular-rate portion.
+* only ``input_tokens`` (Anthropic native) -> cached tokens are *additive*;
+  leave ``input_tokens`` untouched.
+
+``normalize_usage`` maps all of them onto one canonical ``NormalizedUsage``
+shape so billing code needs no vendor knowledge. The known dialects' field
+names do not collide, so a single union parser is safe; a vendor whose fields
+would genuinely conflict needs a dedicated branch here.
+"""
+
+from pydantic.v1 import BaseModel
+
+
+class NormalizedUsage(BaseModel):
+    """Canonical token usage: input_tokens never includes cached tokens."""
+
+    input_tokens: int = 0
+    output_tokens: int = 0
+    cache_read_tokens: int = 0
+    cache_write_tokens: int = 0
+
+
+def parse_token_count(value: object) -> int:
+    """Parse a token count from various formats (int, float, str, bool)."""
+    if isinstance(value, bool):
+        return 0
+    if isinstance(value, int):
+        return max(0, value)
+    if isinstance(value, float):
+        return max(0, int(value))
+    if isinstance(value, str):
+        try:
+            return max(0, int(float(value)))
+        except ValueError:
+            return 0
+    return 0
+
+
+def _first_token_count(usage_data: dict, *fields: str) -> int:
+    """Return the first positive token count among the given fields."""
+    for field in fields:
+        value = parse_token_count(usage_data.get(field, 0))
+        if value > 0:
+            return value
+    return 0
+
+
+def _extract_cache_tokens(usage_data: dict) -> tuple[int, int]:
+    """Pull (cache_read, cache_write) across all known dialects.
+
+    Precedence (highest first), independent for reads and writes:
+
+    * Anthropic top-level: ``cache_read_input_tokens`` /
+      ``cache_creation_input_tokens``.
+    * Nested ``prompt_tokens_details``: ``cached_tokens`` for reads;
+      ``cache_creation_tokens`` (litellm) or ``cache_write_tokens``
+      (OpenRouter) for writes.
+    * DeepSeek: ``prompt_cache_hit_tokens`` for reads (no write concept).
+    """
+    cache_read = parse_token_count(usage_data.get("cache_read_input_tokens", 0))
+    cache_write = parse_token_count(usage_data.get("cache_creation_input_tokens", 0))
+
+    prompt_details = usage_data.get("prompt_tokens_details")
+    if isinstance(prompt_details, dict):
+        if not cache_read:
+            cache_read = parse_token_count(prompt_details.get("cached_tokens", 0))
+        if not cache_write:
+            cache_write = _first_token_count(
+                prompt_details, "cache_creation_tokens", "cache_write_tokens"
+            )
+
+    if not cache_read:
+        # DeepSeek: prompt_tokens = prompt_cache_hit_tokens + prompt_cache_miss_tokens
+        cache_read = parse_token_count(usage_data.get("prompt_cache_hit_tokens", 0))
+
+    return cache_read, cache_write
+
+
+def normalize_usage(usage_data: object) -> NormalizedUsage | None:
+    """Map a vendor usage dict onto the canonical shape, or None if absent.
+
+    Cached reads and writes are subtracted from the input count exactly once,
+    only for dialects that report a ``prompt_tokens`` grand total that already
+    includes them (OpenAI family, DeepSeek, OpenRouter, litellm). Anthropic
+    native reports them additively under ``input_tokens`` and is left untouched.
+    """
+    if not isinstance(usage_data, dict):
+        return None
+
+    output_tokens = _first_token_count(
+        usage_data, "completion_tokens", "output_tokens"
+    )
+    cache_read, cache_write = _extract_cache_tokens(usage_data)
+
+    # ``prompt_tokens`` is the inclusive grand total; ``input_tokens`` (Anthropic
+    # native) excludes cached tokens. The field chosen decides whether to subtract.
+    if "prompt_tokens" in usage_data:
+        input_tokens = parse_token_count(usage_data.get("prompt_tokens", 0))
+        input_tokens = max(0, input_tokens - cache_read - cache_write)
+    else:
+        input_tokens = parse_token_count(usage_data.get("input_tokens", 0))
+
+    return NormalizedUsage(
+        input_tokens=input_tokens,
+        output_tokens=output_tokens,
+        cache_read_tokens=cache_read,
+        cache_write_tokens=cache_write,
+    )

routstr/proxy.py

@@ -1,4 +1,5 @@
 import asyncio
+import inspect
 import json
 from typing import Any
 
@@ -7,7 +8,13 @@ from fastapi.responses import Response, StreamingResponse
 from sqlmodel import select
 
 from .algorithm import create_model_mappings
-from .auth import pay_for_request, revert_pay_for_request, validate_bearer_key
+from .auth import (
+    ReservationSnapshot,
+    get_reservation_snapshot,
+    pay_for_request,
+    revert_pay_for_request,
+    validate_bearer_key,
+)
 from .core import get_logger
 from .core.db import (
     ApiKey,
@@ -24,10 +31,12 @@ from .payment.helpers import (
     calculate_discounted_max_cost,
     check_token_balance,
     create_error_response,
+    create_upstream_error_response,
     get_max_cost_for_model,
 )
 from .payment.models import Model
 from .upstream import BaseUpstreamProvider
+from .upstream.ehbp import forward_ehbp_request, forward_ehbp_x_cashu_request
 from .upstream.helpers import init_upstreams
 from .upstream.request_correction import correct_request, extract_error_message
 
@@ -35,13 +44,19 @@ logger = get_logger(__name__)
 proxy_router = APIRouter()
 
 _upstreams: list[BaseUpstreamProvider] = []
-_model_instances: dict[str, Model] = {}  # All aliases -> Model
 _provider_map: dict[
-    str, list[BaseUpstreamProvider]
-] = {}  # All aliases -> List[Provider]
+    str, list[tuple[Model, BaseUpstreamProvider]]
+] = {}  # All aliases -> sorted [(candidate Model, its Provider)]
 _unique_models: dict[str, Model] = {}  # Unique model.id -> Model (no duplicates)
 
 
+async def _finish_read_transaction(session: AsyncSession) -> None:
+    """Release a read transaction without assuming a particular session mock."""
+    commit_result = session.commit()
+    if inspect.isawaitable(commit_result):
+        await commit_result
+
+
 async def initialize_upstreams() -> None:
     """Initialize upstream providers from database during application startup."""
     global _upstreams
@@ -70,32 +85,44 @@ def get_upstreams() -> list[BaseUpstreamProvider]:
     return _upstreams
 
 
-def get_model_instance(model_id: str) -> Model | None:
-    """Get Model instance by ID from global cache."""
+def get_candidates(
+    model_id: str,
+) -> list[tuple[Model, BaseUpstreamProvider]] | None:
+    """Get the sorted (model, provider) candidate list for a model ID.
+
+    Each provider is paired with its own model for the alias, so routing can
+    forward and bill the candidate that actually serves. Version suffixes
+    (e.g. ``-20251222``) are stripped as a retry when the exact ID is
+    unknown, since upstreams may return a specific version of a base model
+    we track.
+    """
     if not model_id:
         return None
 
     model_id_lower = model_id.lower()
-    # Try exact match first
-    if model := _model_instances.get(model_id_lower):
-        return model
+    if candidates := _provider_map.get(model_id_lower):
+        return candidates
 
-    # Try stripping common version suffixes (e.g., -20251222)
-    # This handles cases where upstream returns a specific version
-    # but we only track the base model name.
     import re
 
     base_model_id = re.sub(r"-\d{8}$", "", model_id_lower)
     if base_model_id != model_id_lower:
-        if model := _model_instances.get(base_model_id):
-            return model
+        if candidates := _provider_map.get(base_model_id):
+            return candidates
 
     return None
 
 
+def get_model_instance(model_id: str) -> Model | None:
+    """Get the best-ranked Model instance for a model ID."""
+    candidates = get_candidates(model_id)
+    return candidates[0][0] if candidates else None
+
+
 def get_provider_for_model(model_id: str) -> list[BaseUpstreamProvider] | None:
-    """Get UpstreamProvider list for model ID from global cache."""
-    return _provider_map.get(model_id.lower())
+    """Get the sorted UpstreamProvider list for a model ID."""
+    candidates = get_candidates(model_id)
+    return [provider for _, provider in candidates] if candidates else None
 
 
 def get_unique_models() -> list[Model]:
@@ -103,11 +130,39 @@ def get_unique_models() -> list[Model]:
     return list(_unique_models.values())
 
 
+def _is_tinfoil_attestation_path(path: str) -> bool:
+    """Return True for exact Tinfoil attestation routes, with optional slash."""
+    return path in {
+        "attestation",
+        "attestation/",
+        "tee/attestation",
+        "tee/attestation/",
+    }
+
+
+def _select_unauthenticated_get_upstreams(
+    path: str, upstreams: list[BaseUpstreamProvider]
+) -> list[BaseUpstreamProvider]:
+    """Select upstream candidates for unauthenticated GET bypass paths.
+
+    Tinfoil attestation endpoints are provider-specific. Trying every enabled
+    upstream can return an unrelated provider's 404 before Tinfoil is reached,
+    so route those paths only to Tinfoil providers.
+    """
+    if _is_tinfoil_attestation_path(path):
+        return [
+            upstream
+            for upstream in upstreams
+            if getattr(upstream, "provider_type", None) == "tinfoil"
+        ]
+    return upstreams
+
+
 async def refresh_model_maps() -> None:
     """Refresh global model and provider maps using the cost-based algorithm."""
     from sqlalchemy.orm import selectinload
 
-    global _model_instances, _provider_map, _unique_models
+    global _provider_map, _unique_models
 
     async with create_session() as session:
         # Fetch all providers with their models in a single logical operation
@@ -117,22 +172,23 @@ async def refresh_model_maps() -> None:
         result = await session.exec(query)
         provider_rows = result.all()
 
-    overrides_by_id: dict[str, tuple[ModelRow, float]] = {}
-    disabled_model_ids: set[str] = set()
+    overrides_by_key: dict[tuple[str, int], tuple[ModelRow, float]] = {}
+    disabled_model_keys: set[tuple[str, int]] = set()
 
     for provider in provider_rows:
         if not provider.enabled:
             continue
         for model in provider.models:
+            model_key = (model.id.lower(), model.upstream_provider_id)
             if model.enabled:
-                overrides_by_id[model.id] = (model, provider.provider_fee)
+                overrides_by_key[model_key] = (model, provider.provider_fee)
             else:
-                disabled_model_ids.add(model.id)
+                disabled_model_keys.add(model_key)
 
-    _model_instances, _provider_map, _unique_models = create_model_mappings(
+    _, _provider_map, _unique_models = create_model_mappings(
         upstreams=_upstreams,
-        overrides_by_id=overrides_by_id,
-        disabled_model_ids=disabled_model_ids,
+        overrides_by_key=overrides_by_key,
+        disabled_model_keys=disabled_model_keys,
     )
 
 
@@ -165,12 +221,27 @@ _API_PATH_PREFIXES = (
     "moderations",
     "providers",
     "tee/",
+    "attestation",
 )
 
 
 @proxy_router.api_route("/{path:path}", methods=["GET", "POST"], response_model=None)
 async def proxy(
     request: Request, path: str, session: AsyncSession = Depends(get_session)
+) -> Response | StreamingResponse:
+    """Run proxy setup in a short request session, never across response streaming."""
+    try:
+        return await _proxy(request, path, session)
+    finally:
+        # FastAPI yield dependencies normally close after the response body is
+        # sent. Close explicitly so a long stream cannot retain DB resources.
+        close_result = session.close()
+        if inspect.isawaitable(close_result):
+            await close_result
+
+
+async def _proxy(
+    request: Request, path: str, session: AsyncSession
 ) -> Response | StreamingResponse:
     # GET requests must hit a known API prefix; otherwise return a 404 (HTML
     # for browsers, JSON for API clients). POST requests are always forwarded
@@ -183,20 +254,54 @@ async def proxy(
 
     is_responses_api = path.startswith("v1/responses") or path.startswith("responses")
     request_body = await request.body()
-    request_body_dict = parse_request_body_json(request_body, path)
 
-    # /tee/* GET requests (e.g. attestation) don't map to models — just
-    # forward to all enabled upstreams without model/cost/auth lookups.
-    if request.method == "GET" and path.startswith("tee/"):
-        all_upstreams = _upstreams
+    # EHBP (Encrypted HTTP Body Protocol) requests carry an Ehbp-Encapsulated-Key
+    # header and a binary HPKE-sealed body. The proxy cannot parse the body to
+    # extract the model id, so the SDK sends it in X-Routstr-Model. Forward the
+    # raw encrypted body to the upstream's /private/ endpoint and stream the
+    # encrypted response back untouched — the SDK's SecureClient decrypts it.
+    is_ehbp = "ehbp-encapsulated-key" in headers
+    if is_ehbp:
+        request_body_dict = {}
+        model_id = headers.get("x-routstr-model", "")
+        if not model_id:
+            return create_error_response(
+                "invalid_request",
+                "EHBP request missing X-Routstr-Model header",
+                400,
+                request=request,
+            )
+    else:
+        request_body_dict = parse_request_body_json(request_body, path)
+        if is_responses_api:
+            model_id = extract_model_from_responses_request(request_body_dict)
+        else:
+            model_id = request_body_dict.get("model", "unknown")
+
+    # Exact Tinfoil attestation GET routes don't map to models — forward
+    # without model/cost/auth lookups. Do not prefix-match here: paths such as
+    # /attestationjunk must continue through normal authentication.
+    if request.method == "GET" and _is_tinfoil_attestation_path(path):
+        selected_upstreams = _select_unauthenticated_get_upstreams(path, _upstreams)
+        if not selected_upstreams:
+            return create_error_response(
+                "upstream_error",
+                "No upstream available for unauthenticated GET path",
+                502,
+                request=request,
+            )
+
         last_error_response = None
-        for i, upstream in enumerate(all_upstreams):
+        for i, upstream in enumerate(selected_upstreams):
             try:
                 headers = upstream.prepare_headers(dict(request.headers))
                 response = await upstream.forward_get_request(request, path, headers)
-                if response.status_code in [502, 429] and i < len(all_upstreams) - 1:
+                if (
+                    response.status_code in [502, 429]
+                    and i < len(selected_upstreams) - 1
+                ):
                     logger.warning(
-                        "Upstream %s returned %s for tee GET %s, trying next",
+                        "Upstream %s returned %s for unauthenticated GET %s, trying next",
                         upstream.provider_type,
                         response.status_code,
                         path,
@@ -205,44 +310,43 @@ async def proxy(
                 return response
             except UpstreamError as e:
                 logger.warning(
-                    "Upstream %s failed for tee GET %s: %s",
+                    "Upstream %s failed for unauthenticated GET %s: %s",
                     upstream.provider_type,
                     path,
                     e,
                 )
-                if i == len(all_upstreams) - 1:
-                    last_error_response = create_error_response(
-                        "upstream_error", str(e), 502, request=request
-                    )
+                if i == len(selected_upstreams) - 1:
+                    last_error_response = create_upstream_error_response(e, request)
                 continue
         return last_error_response or create_error_response(
             "upstream_error", "All upstreams failed", 502, request=request
         )
 
-    if is_responses_api:
-        model_id = extract_model_from_responses_request(request_body_dict)
-    else:
-        model_id = request_body_dict.get("model", "unknown")
+    candidates = get_candidates(model_id)
 
-    model_obj = get_model_instance(model_id)
-
-    if not model_obj:
+    if not candidates:
         return create_error_response(
             "invalid_model", f"Model '{model_id}' not found", 400, request=request
         )
 
-    upstreams = get_provider_for_model(model_id)
-    if not upstreams:
-        return create_error_response(
-            "invalid_model",
-            f"No provider found for model '{model_id}'",
-            400,
-            request=request,
-        )
+    if is_ehbp:
+        candidates = [
+            (model, upstream)
+            for model, upstream in candidates
+            if upstream.supports_ehbp
+        ]
+        if not candidates:
+            return create_error_response(
+                "unsupported_request",
+                f"No EHBP-capable provider found for model '{model_id}'",
+                400,
+                request=request,
+            )
 
-    # todo figure out cost calculation since fallback provider is usually not the same price
-    # Use first provider for initial checks/cost calculation
-    # primary_upstream = upstreams[0]
+    # Reserve/max-cost checks use the best-ranked candidate; the failover loop
+    # below rebinds (model_obj, upstream) per candidate so forwarding and
+    # settlement always use the model of the provider actually being tried.
+    model_obj = candidates[0][0]
 
     _max_cost_for_model = await get_max_cost_for_model(
         model=model_id, session=session, model_obj=model_obj
@@ -257,9 +361,25 @@ async def proxy(
 
     if x_cashu := headers.get("x-cashu", None):
         last_error = None
-        for i, upstream in enumerate(upstreams):
+        for i, (model_obj, upstream) in enumerate(candidates):
             try:
-                if is_responses_api:
+                if is_ehbp:
+                    if not upstream.supports_ehbp:
+                        logger.warning(
+                            "Upstream %s does not support EHBP for model=%s",
+                            upstream.provider_type,
+                            model_id,
+                        )
+                        continue
+                    return await forward_ehbp_x_cashu_request(
+                        request=request,
+                        x_cashu_token=x_cashu,
+                        path=path,
+                        max_cost_for_model=max_cost_for_model,
+                        model_obj=model_obj,
+                        upstream=upstream,
+                    )
+                elif is_responses_api:
                     return await upstream.handle_x_cashu_responses(
                         request, x_cashu, path, max_cost_for_model, model_obj
                     )
@@ -279,15 +399,14 @@ async def proxy(
                         "status_code": e.status_code,
                     },
                 )
-                if i == len(upstreams) - 1:
+                if i == len(candidates) - 1:
                     last_error = e
                 continue
 
+        if last_error is not None:
+            return create_upstream_error_response(last_error, request)
         return create_error_response(
-            "upstream_error",
-            str(last_error) if last_error else "All upstreams failed",
-            502,
-            request=request,
+            "upstream_error", "All upstreams failed", 502, request=request
         )
 
     elif auth := headers.get("authorization", None):
@@ -307,12 +426,12 @@ async def proxy(
         logger.debug("Processing unauthenticated GET request", extra={"path": path})
 
         last_error_response = None
-        for i, upstream in enumerate(upstreams):
+        for i, (_, upstream) in enumerate(candidates):
             try:
                 headers = upstream.prepare_headers(dict(request.headers))
                 response = await upstream.forward_get_request(request, path, headers)
 
-                if response.status_code in [502, 429] and i < len(upstreams) - 1:
+                if response.status_code in [502, 429] and i < len(candidates) - 1:
                     error_message = ""
                     try:
                         if hasattr(response, "body"):
@@ -342,30 +461,86 @@ async def proxy(
                 return response
             except UpstreamError as e:
                 logger.warning(f"Upstream {upstream.provider_type} failed (GET): {e}")
-                if i == len(upstreams) - 1:
-                    last_error_response = create_error_response(
-                        "upstream_error", str(e), 502, request=request
-                    )
+                if i == len(candidates) - 1:
+                    last_error_response = create_upstream_error_response(e, request)
                 continue
         return last_error_response or create_error_response(
             "upstream_error", "All upstreams failed", 502, request=request
         )
 
-    if request_body_dict:
+    reservation_snapshot: ReservationSnapshot | None = None
+    if is_ehbp or request_body_dict:
         await pay_for_request(key, max_cost_for_model, session)
+        reservation_snapshot = await get_reservation_snapshot(key, session)
+        # Snapshot validation performs SELECTs after pay_for_request commits.
+        # End that read transaction before waiting on upstream response headers.
+        await _finish_read_transaction(session)
 
     # Tracks request params already removed in response to upstream rejections,
     # shared across providers so a stripped param stays stripped on failover and
     # the reactive retry can never loop unboundedly.
     already_stripped: set[str] = set()
 
-    for i, upstream in enumerate(upstreams):
+    for i, (model_obj, upstream) in enumerate(candidates):
+        if i > 0 and request_body_dict:
+            # The reservation was sized to the previous candidate's envelope;
+            # settlement bills the serving candidate, so a pricier fallback
+            # must be re-reserved at its own max cost before it is tried. A
+            # candidate whose envelope the key cannot cover is rejected, just
+            # as it would be had it been ranked first.
+            candidate_max = await get_max_cost_for_model(
+                model=model_id, session=session, model_obj=model_obj
+            )
+            candidate_max = await calculate_discounted_max_cost(
+                candidate_max, request_body_dict, model_obj=model_obj
+            )
+            candidate_max = max(candidate_max, settings.min_request_msat)
+            if candidate_max > max_cost_for_model:
+                await revert_pay_for_request(
+                    key, session, max_cost_for_model, reservation_snapshot
+                )
+                try:
+                    await pay_for_request(key, candidate_max, session)
+                except HTTPException:
+                    if i == len(candidates) - 1:
+                        raise
+                    await pay_for_request(key, max_cost_for_model, session)
+                    reservation_snapshot = await get_reservation_snapshot(key, session)
+                    await _finish_read_transaction(session)
+                    continue
+                reservation_snapshot = await get_reservation_snapshot(key, session)
+                await _finish_read_transaction(session)
+                max_cost_for_model = candidate_max
+
         headers = upstream.prepare_headers(dict(request.headers))
 
         try:
             while True:
                 try:
-                    if is_responses_api:
+                    if is_ehbp:
+                        if not upstream.supports_ehbp:
+                            logger.warning(
+                                "Upstream %s does not support EHBP for model=%s",
+                                upstream.provider_type,
+                                model_id,
+                            )
+                            raise UpstreamError(
+                                f"Provider {upstream.provider_type} does not support EHBP",
+                                status_code=400,
+                            )
+                        response = await forward_ehbp_request(
+                            request=request,
+                            path=path,
+                            headers=headers,
+                            request_body=request_body,
+                            upstream=upstream,
+                            key=key,
+                            max_cost_for_model=max_cost_for_model,
+                            session=session,
+                            model_obj=model_obj,
+                            reservation_snapshot=reservation_snapshot,
+                        )
+                    elif is_responses_api:
                         response = await upstream.forward_responses_request(
                             request,
                             path,
@@ -375,6 +550,7 @@ async def proxy(
                             max_cost_for_model,
                             session,
                             model_obj,
+                            reservation_snapshot,
                         )
                     else:
                         response = await upstream.forward_request(
@@ -386,6 +562,7 @@ async def proxy(
                             max_cost_for_model,
                             session,
                             model_obj,
+                            reservation_snapshot,
                         )
                 except UpstreamError:
                     # Let the outer UpstreamError handler manage retry/revert
@@ -402,7 +579,9 @@ async def proxy(
                             "max_cost_for_model": max_cost_for_model,
                         },
                     )
-                    await revert_pay_for_request(key, session, max_cost_for_model)
+                    await revert_pay_for_request(
+                        key, session, max_cost_for_model, reservation_snapshot
+                    )
                     raise
 
                 # Reactive recovery: some models reject one specific request
@@ -410,7 +589,7 @@ async def proxy(
                 # When the upstream 400s naming such a param, strip it from the
                 # body and retry the SAME upstream. ``already_stripped`` bounds
                 # this to one retry per distinct param so it always terminates.
-                if response.status_code == 400:
+                if response.status_code == 400 and not is_ehbp:
                     correction = correct_request(
                         request_body,
                         extract_error_message(response),
@@ -438,7 +617,7 @@ async def proxy(
             if response.status_code != 200:
                 # Check if we should retry (502 Upstream Error or 429 Rate Limit)
                 should_retry = response.status_code in [502, 429, 400, 401, 403, 404]
-                if should_retry and i < len(upstreams) - 1:
+                if should_retry and i < len(candidates) - 1:
                     error_message = ""
                     try:
                         if hasattr(response, "body"):
@@ -471,7 +650,9 @@ async def proxy(
                     continue
 
                 # 4xx error (user error), or other non-retryable error, or last provider failed
-                await revert_pay_for_request(key, session, max_cost_for_model)
+                await revert_pay_for_request(
+                    key, session, max_cost_for_model, reservation_snapshot
+                )
                 logger.warning(
                     "Upstream request failed, revert payment "
                     "(provider=%s model=%s status=%s path=%s)",
@@ -503,8 +684,10 @@ async def proxy(
                     "max_cost_for_model": max_cost_for_model,
                 },
             )
-            await asyncio.shield(
-                revert_pay_for_request(key, session, max_cost_for_model)
+            # The cancellation has been caught, so complete exact cleanup in
+            # this task before the request-scoped session can be torn down.
+            await revert_pay_for_request(
+                key, session, max_cost_for_model, reservation_snapshot
             )
             raise
 
@@ -518,16 +701,16 @@ async def proxy(
                     "provider": upstream.provider_type,
                     "model": model_id,
                     "status_code": e.status_code,
-                    "retry": i < len(upstreams) - 1,
+                    "retry": i < len(candidates) - 1,
                 },
             )
 
             # If this was the last provider
-            if i == len(upstreams) - 1:
-                await revert_pay_for_request(key, session, max_cost_for_model)
-                return create_error_response(
-                    "upstream_error", str(e), 502, request=request
+            if i == len(candidates) - 1:
+                await revert_pay_for_request(
+                    key, session, max_cost_for_model, reservation_snapshot
                 )
+                return create_upstream_error_response(e, request)
 
             # Otherwise loop continues to next provider
             continue

routstr/upstream/__init__.py

@@ -11,6 +11,7 @@ from .openrouter import OpenRouterUpstreamProvider
 from .perplexity import PerplexityUpstreamProvider
 from .ppqai import PPQAIUpstreamProvider
 from .routstr import RoutstrUpstreamProvider
+from .tinfoil import TinfoilUpstreamProvider
 from .xai import XAIUpstreamProvider
 
 upstream_provider_classes: list[type[BaseUpstreamProvider]] = [
@@ -26,6 +27,7 @@ upstream_provider_classes: list[type[BaseUpstreamProvider]] = [
     PerplexityUpstreamProvider,
     PPQAIUpstreamProvider,
     RoutstrUpstreamProvider,
+    TinfoilUpstreamProvider,
     XAIUpstreamProvider,
 ]
 """List of all upstream classes"""

routstr/upstream/anthropic.py

@@ -24,7 +24,7 @@ class AnthropicUpstreamProvider(BaseUpstreamProvider):
         )
 
     @classmethod
-    def from_db_row(
+    def _build_from_row(
         cls, provider_row: "UpstreamProviderRow"
     ) -> "AnthropicUpstreamProvider":
         return cls(

routstr/upstream/auto_topup.py

@@ -4,7 +4,14 @@ import json
 from sqlmodel import select
 
 from ..core import get_logger
-from ..core.db import UpstreamProviderRow, create_session
+from ..core.db import (
+    CashuTransaction,
+    UpstreamProviderRow,
+    create_session,
+)
+from ..core.db import (
+    store_cashu_transaction_with_retry as store_cashu_transaction,
+)
 from ..wallet import send_token
 from .routstr import RoutstrUpstreamProvider
 
@@ -97,6 +104,8 @@ async def _check_and_topup(row: UpstreamProviderRow) -> None:
 
     # Instantiate provider and check balance
     provider = RoutstrUpstreamProvider.from_db_row(row)
+    if provider is None:
+        return
     balance = await provider.get_balance()
 
     if balance is None:
@@ -121,7 +130,6 @@ async def _check_and_topup(row: UpstreamProviderRow) -> None:
         },
     )
 
-    print(amount, mint_url)
     try:
         token = await send_token(amount, "sat", mint_url)
     except Exception as e:
@@ -136,6 +144,23 @@ async def _check_and_topup(row: UpstreamProviderRow) -> None:
         )
         return
 
+    try:
+        await store_cashu_transaction(
+            token=token,
+            amount=amount,
+            unit="sat",
+            mint_url=mint_url,
+            typ="out",
+            collected=False,
+            source="auto_topup",
+        )
+    except Exception:
+        logger.critical(
+            "Aborting auto top-up because its cashu token could not be persisted",
+            extra={"provider_id": row.id, "mint_url": mint_url},
+        )
+        return
+
     result = await provider.topup(token)
 
     if "error" in result:
@@ -147,6 +172,26 @@ async def _check_and_topup(row: UpstreamProviderRow) -> None:
             },
         )
     else:
+        async with create_session() as session:
+            transaction = (
+                await session.exec(
+                    select(CashuTransaction).where(
+                        CashuTransaction.token == token,
+                        CashuTransaction.type == "out",
+                        CashuTransaction.source == "auto_topup",
+                    )
+                )
+            ).first()
+            if transaction is None:
+                logger.critical(
+                    "Completed auto top-up transaction is missing from the database",
+                    extra={"provider_id": row.id, "mint_url": mint_url},
+                )
+            else:
+                transaction.collected = True
+                session.add(transaction)
+                await session.commit()
+
         logger.info(
             "Auto top-up completed successfully",
             extra={

routstr/upstream/azure.py

@@ -38,7 +38,7 @@ class AzureUpstreamProvider(BaseUpstreamProvider):
         self.api_version = api_version
 
     @classmethod
-    def from_db_row(
+    def _build_from_row(
         cls, provider_row: "UpstreamProviderRow"
     ) -> "AzureUpstreamProvider | None":
         if not provider_row.api_version:

routstr/upstream/cache_breakpoints.py

@@ -0,0 +1,157 @@
+"""Inject explicit prompt-cache breakpoints into OpenAI-shaped requests.
+
+Some upstreams cache *explicitly*: the request must carry
+``cache_control: {"type": "ephemeral"}`` markers on the content blocks that
+should be cached. Two model families use this identical wire format:
+
+* **Anthropic Claude** — direct or via OpenRouter's ``anthropic/*`` models.
+* **Alibaba's explicit-cache models on OpenRouter** — ``qwen/qwen3-max``,
+  ``qwen/qwen-plus``, ``qwen/qwen3.6-plus``, ``qwen/qwen3-coder-plus``,
+  ``qwen/qwen3-coder-flash`` and ``deepseek/deepseek-v3.2`` — which OpenRouter
+  documents as using "the same syntax as Anthropic explicit caching".
+
+Every other provider routstr proxies (OpenAI, Azure, xAI/Grok, Groq, Moonshot,
+default DeepSeek, Gemini implicit, Fireworks) caches *automatically* and needs
+no markers — they are left untouched.
+
+A client that doesn't know it is talking to one of these models *through*
+routstr (e.g. an OpenAI-compatible coding agent pointed at a routstr URL) never
+emits the markers — it only adds them when it recognises the provider as
+OpenRouter. So caching silently never engages over routstr even though the same
+client caches fine talking to OpenRouter directly.
+
+This module restores caching by stamping the standard breakpoints onto the
+forwarded body — the system prompt, the last tool, and the last conversation
+message (the format allows up to four; we use three, matching the common
+agent convention) — but only when the client supplied none of its own, so
+explicit client control always wins. The caller is responsible for only
+applying this toward an upstream that accepts the markers (OpenRouter /
+Anthropic), so they never leak to a provider that would reject them.
+"""
+
+from __future__ import annotations
+
+from typing import Any
+
+# The single ephemeral marker stamped onto each chosen breakpoint. A 5-minute
+# TTL (the default for ``ephemeral``) — deliberately not the 1h tier, which
+# carries a higher cache-write premium and should stay opt-in.
+EPHEMERAL_CACHE_CONTROL: dict[str, str] = {"type": "ephemeral"}
+
+# Alibaba's explicit-cache models on OpenRouter. Matched as substrings of the
+# model id (any spelling routstr carries). Snapshot endpoints that OpenRouter
+# documents as *not* supporting explicit caching (e.g. ``qwen3.5-plus-02-15``)
+# are different families and deliberately absent from this list.
+_ALIBABA_EXPLICIT_CACHE_SLUGS: tuple[str, ...] = (
+    "qwen3-max",
+    "qwen-plus",
+    "qwen3.6-plus",
+    "qwen3-coder-plus",
+    "qwen3-coder-flash",
+    "deepseek-v3.2",
+)
+
+
+def is_explicit_cache_model(model_id: str | None, *fallbacks: str | None) -> bool:
+    """True when the target model uses the explicit ``cache_control`` dialect.
+
+    Covers the Claude family (broadly — every Claude model supports it) and
+    Alibaba's documented explicit-cache models, across the id spellings routstr
+    carries: the OpenRouter id (``anthropic/claude-...``, ``qwen/qwen3-max``),
+    the bare upstream id, and any forwarded/canonical alias.
+    """
+    for candidate in (model_id, *fallbacks):
+        if not candidate:
+            continue
+        lowered = candidate.lower()
+        if "claude" in lowered or "anthropic/" in lowered:
+            return True
+        if any(slug in lowered for slug in _ALIBABA_EXPLICIT_CACHE_SLUGS):
+            return True
+    return False
+
+
+def _has_cache_control(obj: Any) -> bool:
+    """Recursively detect any client-supplied ``cache_control`` marker."""
+    if isinstance(obj, dict):
+        if "cache_control" in obj:
+            return True
+        return any(_has_cache_control(v) for v in obj.values())
+    if isinstance(obj, list):
+        return any(_has_cache_control(v) for v in obj)
+    return False
+
+
+def body_has_cache_control(data: dict) -> bool:
+    """True when the request already carries cache_control on messages/tools."""
+    return _has_cache_control(data.get("messages")) or _has_cache_control(
+        data.get("tools")
+    )
+
+
+def _stamp_text_content(message: dict) -> bool:
+    """Add the ephemeral marker to a message's last text block.
+
+    A string content is promoted to the array form Anthropic requires for
+    cache markers; an existing array gets the marker on its last text part.
+    Returns True when a marker was placed.
+    """
+    content = message.get("content")
+    if isinstance(content, str):
+        if not content:
+            return False
+        message["content"] = [
+            {
+                "type": "text",
+                "text": content,
+                "cache_control": dict(EPHEMERAL_CACHE_CONTROL),
+            }
+        ]
+        return True
+    if isinstance(content, list):
+        for part in reversed(content):
+            if isinstance(part, dict) and part.get("type") == "text":
+                part["cache_control"] = dict(EPHEMERAL_CACHE_CONTROL)
+                return True
+    return False
+
+
+def _stamp_system_prompt(messages: list) -> None:
+    for message in messages:
+        if isinstance(message, dict) and message.get("role") in (
+            "system",
+            "developer",
+        ):
+            _stamp_text_content(message)
+            return
+
+
+def _stamp_last_tool(tools: Any) -> None:
+    if isinstance(tools, list) and tools and isinstance(tools[-1], dict):
+        tools[-1]["cache_control"] = dict(EPHEMERAL_CACHE_CONTROL)
+
+
+def _stamp_last_conversation_message(messages: list) -> None:
+    for message in reversed(messages):
+        if isinstance(message, dict) and message.get("role") in ("user", "assistant"):
+            if _stamp_text_content(message):
+                return
+
+
+def inject_anthropic_cache_breakpoints(data: dict) -> bool:
+    """Stamp ephemeral cache breakpoints onto an OpenAI-shaped chat body.
+
+    Mutates ``data`` in place (the established convention in
+    ``prepare_request_body``) and returns True when anything changed. No-ops
+    when the body isn't chat-shaped or the client already set cache_control.
+    """
+    messages = data.get("messages")
+    if not isinstance(messages, list) or not messages:
+        return False
+    if body_has_cache_control(data):
+        return False
+
+    _stamp_system_prompt(messages)
+    _stamp_last_tool(data.get("tools"))
+    _stamp_last_conversation_message(messages)
+    return True

routstr/upstream/deepseek_v4_pricing_shim.py

@@ -0,0 +1,73 @@
+"""TEMPORARY: local DeepSeek V4 pricing shim.
+
+litellm's bundled cost map does not yet ship ``deepseek-v4-flash`` /
+``deepseek-v4-pro``. Without an entry, ``backfill_cache_pricing`` cannot find a
+``cache_read_input_token_cost`` and cache reads fall back to the full input
+rate — a large overcharge on cache hits (DeepSeek V4 hits are ~0.008-0.02x
+input, i.e. cached tokens cost 50-120x less than regular input).
+
+This module injects the missing entries into ``litellm.model_cost`` at startup
+so the existing backfill path resolves them. Rates mirror the canonical
+``deepseek`` provider entries now in litellm's ``model_prices`` map
+(``input_cost_per_token`` is the cache-*miss* rate;
+``cache_read_input_token_cost`` is the cache-*hit* rate), sourced from
+https://api-docs.deepseek.com/quick_start/pricing via
+https://github.com/BerriAI/litellm/pull/26380 (issue
+https://github.com/BerriAI/litellm/issues/30430).
+
+=== REMOVAL (once litellm ships these models) ===
+Delete this file and the single ``register_deepseek_v4_pricing()`` call in
+``routstr/core/main.py``. Nothing else depends on it. Entries are only added
+when absent, so a stale shim is harmless after upstream lands — but remove it.
+"""
+
+import litellm
+
+from ..core import get_logger
+
+logger = get_logger(__name__)
+
+# USD per token. Mirrors the canonical ``deepseek`` provider entries in
+# litellm's model_prices map (source: DeepSeek API pricing docs). Keep these in
+# sync with ``litellm.model_cost["deepseek/deepseek-v4-*"]``.
+_DEEPSEEK_V4_RATES: dict[str, dict[str, float]] = {
+    "deepseek-v4-flash": {
+        "input_cost_per_token": 1.4e-07,
+        "output_cost_per_token": 2.8e-07,
+        "cache_read_input_token_cost": 2.8e-09,
+        "cache_creation_input_token_cost": 0.0,
+        "input_cost_per_token_cache_hit": 2.8e-09,
+    },
+    "deepseek-v4-pro": {
+        "input_cost_per_token": 4.35e-07,
+        "output_cost_per_token": 8.7e-07,
+        "cache_read_input_token_cost": 3.625e-09,
+        "cache_creation_input_token_cost": 0.0,
+        "input_cost_per_token_cache_hit": 3.625e-09,
+    },
+}
+
+
+def register_deepseek_v4_pricing() -> None:
+    """Inject DeepSeek V4 pricing into ``litellm.model_cost`` if absent.
+
+    Idempotent and non-destructive: a key already present in the cost map
+    (e.g. once litellm ships it) is left untouched. Registers both the bare
+    (``deepseek-v4-flash``) and prefixed (``deepseek/deepseek-v4-flash``)
+    spellings since ``backfill_cache_pricing`` tries both.
+    """
+    added = []
+    for bare, rates in _DEEPSEEK_V4_RATES.items():
+        for key in (bare, f"deepseek/{bare}"):
+            if key in litellm.model_cost:
+                continue
+            entry: dict[str, object] = dict(rates)
+            entry["litellm_provider"] = "deepseek"
+            entry["mode"] = "chat"
+            litellm.model_cost[key] = entry
+            added.append(key)
+    if added:
+        logger.info(
+            "Registered temporary DeepSeek V4 pricing shim",
+            extra={"models": added},
+        )

routstr/upstream/fireworks.py

@@ -20,7 +20,7 @@ class FireworksUpstreamProvider(BaseUpstreamProvider):
         )
 
     @classmethod
-    def from_db_row(
+    def _build_from_row(
         cls, provider_row: "UpstreamProviderRow"
     ) -> "FireworksUpstreamProvider":
         return cls(

routstr/upstream/gemini.py

@@ -51,7 +51,7 @@ class GeminiUpstreamProvider(BaseUpstreamProvider):
         return self._client
 
     @classmethod
-    def from_db_row(
+    def _build_from_row(
         cls, provider_row: "UpstreamProviderRow"
     ) -> "GeminiUpstreamProvider":
         return cls(

routstr/upstream/generic.py

@@ -5,6 +5,12 @@ from typing import TYPE_CHECKING
 import httpx
 
 from .base import BaseUpstreamProvider
+from .pricing_resolver import (
+    FallbackPricingResolver,
+    ResolvedPricing,
+    _as_float,
+    estimate_context_length,
+)
 
 if TYPE_CHECKING:
     from ..core.db import UpstreamProviderRow
@@ -45,7 +51,7 @@ class GenericUpstreamProvider(BaseUpstreamProvider):
         )
 
     @classmethod
-    def from_db_row(
+    def _build_from_row(
         cls, provider_row: "UpstreamProviderRow"
     ) -> "GenericUpstreamProvider":
         return cls(
@@ -64,6 +70,40 @@ class GenericUpstreamProvider(BaseUpstreamProvider):
             "platform_url": cls.platform_url,
         }
 
+    def _native_pricing(
+        self, model_id: str, model_spec: dict
+    ) -> ResolvedPricing | None:
+        """Read pricing/metadata from Venice's bespoke ``model_spec`` schema.
+
+        Returns ``None`` when the upstream reported no *usable* native price —
+        absent, non-numeric, negative, or both-zero — so the caller falls
+        through to the shared resolution chain instead of fabricating a number
+        or trusting a bogus one. This mirrors the money-safety guards the
+        litellm and OpenRouter rungs already apply: a both-zero price would
+        serve the model free, a negative one would credit the caller, and a
+        non-numeric string would otherwise throw and drop the whole catalog.
+        """
+        pricing_info = model_spec.get("pricing", {})
+        input_usd = _as_float(pricing_info.get("input", {}).get("usd"))
+        output_usd = _as_float(pricing_info.get("output", {}).get("usd"))
+        if input_usd is None or output_usd is None:
+            return None
+        if input_usd < 0 or output_usd < 0 or (input_usd == 0 and output_usd == 0):
+            return None
+
+        capabilities = model_spec.get("capabilities", {})
+        input_modalities = ["text"]
+        if capabilities.get("supportsVision", False):
+            input_modalities.append("image")
+
+        return ResolvedPricing(
+            prompt=input_usd / 1_000_000,
+            completion=output_usd / 1_000_000,
+            context_length=model_spec.get("availableContextTokens"),
+            source="native",
+            input_modalities=input_modalities,
+        )
+
     async def fetch_models(self) -> list[Model]:
         """Fetch models from upstream API using /models endpoint."""
         from ..payment.models import Architecture, Model, Pricing, TopProvider
@@ -78,6 +118,7 @@ class GenericUpstreamProvider(BaseUpstreamProvider):
                 response.raise_for_status()
                 data = response.json()
 
+                resolver = FallbackPricingResolver()
                 models_list = []
                 for model_data in data.get("data", []):
                     model_id = model_data.get("id", "")
@@ -89,41 +130,44 @@ class GenericUpstreamProvider(BaseUpstreamProvider):
                     owned_by = model_data.get("owned_by", "unknown")
                     model_spec = model_data.get("model_spec", {})
 
-                    context_length = 4096
-                    if model_spec.get("availableContextTokens"):
-                        context_length = model_spec["availableContextTokens"]
-                    elif any(
-                        pattern in model_id.lower() for pattern in ["32k", "32000"]
-                    ):
-                        context_length = 32768
-                    elif any(
-                        pattern in model_id.lower() for pattern in ["16k", "16000"]
-                    ):
-                        context_length = 16384
-                    elif any(pattern in model_id.lower() for pattern in ["8k", "8000"]):
-                        context_length = 8192
-                    elif "gpt-4" in model_id.lower():
-                        context_length = 8192
-                    elif "claude" in model_id.lower():
-                        context_length = 200000
+                    resolved = self._native_pricing(model_id, model_spec)
+                    if resolved is None:
+                        resolved = await resolver.resolve(model_id)
 
-                    pricing_info = model_spec.get("pricing", {})
-                    input_pricing = pricing_info.get("input", {})
-                    output_pricing = pricing_info.get("output", {})
+                    if resolved is None:
+                        # Fail closed: never invent a price. Import the model
+                        # disabled with a warning so the operator can price it
+                        # (the admin UI surfaces disabled remote models).
+                        logger.warning(
+                            f"No pricing source resolved for '{model_id}' from "
+                            f"{self.upstream_name}; importing it disabled",
+                            extra={"model_id": model_id, "base_url": self.base_url},
+                        )
+                        resolved = ResolvedPricing(
+                            prompt=0.0,
+                            completion=0.0,
+                            context_length=None,
+                            source="unresolved",
+                        )
+                        enabled = False
+                    else:
+                        enabled = True
 
-                    prompt_price = input_pricing.get("usd", 0.001) / 1000000
-                    completion_price = output_pricing.get("usd", 0.001) / 1000000
+                    # Prefer the source's own modality string (OpenRouter ships
+                    # one, e.g. "text+image->text"); otherwise derive it from the
+                    # captured input/output modalities in the same "in->out" shape
+                    # rather than flattening vision models to "text->text".
+                    modality = resolved.modality or (
+                        f"{'+'.join(resolved.input_modalities)}"
+                        f"->{'+'.join(resolved.output_modalities)}"
+                    )
 
-                    capabilities = model_spec.get("capabilities", {})
-                    input_modalities = ["text"]
-                    output_modalities = ["text"]
-
-                    if capabilities.get("supportsVision", False):
-                        input_modalities.append("image")
-
-                    modality = "text"
-                    if capabilities.get("supportsVision", False):
-                        modality = "text->text"
+                    # A source can carry a price but no context (e.g. a litellm
+                    # entry missing max_input_tokens); fall back to an id-based
+                    # estimate so we never persist a zero-length window.
+                    context_length = resolved.context_length or estimate_context_length(
+                        model_id
+                    )
 
                     spec_name = model_spec.get("name", model_name)
                     description = f"{spec_name}"
@@ -139,30 +183,33 @@ class GenericUpstreamProvider(BaseUpstreamProvider):
                             context_length=context_length,
                             architecture=Architecture(
                                 modality=modality,
-                                input_modalities=input_modalities,
-                                output_modalities=output_modalities,
-                                tokenizer="unknown",
-                                instruct_type=None,
+                                input_modalities=resolved.input_modalities,
+                                output_modalities=resolved.output_modalities,
+                                tokenizer=resolved.tokenizer,
+                                instruct_type=resolved.instruct_type,
                             ),
                             pricing=Pricing(
-                                prompt=prompt_price,
-                                completion=completion_price,
+                                prompt=resolved.prompt,
+                                completion=resolved.completion,
                                 request=0.0,
                                 image=0.0,
                                 web_search=0.0,
                                 internal_reasoning=0.0,
-                                max_prompt_cost=0.001,
-                                max_completion_cost=0.001,
-                                max_cost=0.001,
+                                input_cache_read=resolved.input_cache_read,
+                                input_cache_write=resolved.input_cache_write,
                             ),
                             sats_pricing=None,
                             per_request_limits=None,
                             top_provider=TopProvider(
                                 context_length=context_length,
-                                max_completion_tokens=context_length // 2,
-                                is_moderated=False,
+                                max_completion_tokens=(
+                                    resolved.max_completion_tokens
+                                    if resolved.max_completion_tokens is not None
+                                    else context_length // 2
+                                ),
+                                is_moderated=bool(resolved.is_moderated),
                             ),
-                            enabled=True,
+                            enabled=enabled,
                             upstream_provider_id=None,
                             canonical_slug=None,
                         )

routstr/upstream/groq.py

@@ -20,7 +20,7 @@ class GroqUpstreamProvider(BaseUpstreamProvider):
         )
 
     @classmethod
-    def from_db_row(cls, provider_row: "UpstreamProviderRow") -> "GroqUpstreamProvider":
+    def _build_from_row(cls, provider_row: "UpstreamProviderRow") -> "GroqUpstreamProvider":
         return cls(
             api_key=provider_row.api_key,
             provider_fee=provider_row.provider_fee,

routstr/upstream/helpers.py

@@ -12,6 +12,7 @@ from sqlmodel import select
 
 from ..core import get_logger
 from ..core.db import AsyncSession, ModelRow, UpstreamProviderRow, create_session
+from ..core.provider_slugs import allocate_unique_provider_slug
 from ..payment.models import Model
 from .base import BaseUpstreamProvider
 
@@ -93,12 +94,10 @@ async def get_all_models_with_overrides(
         provider_result = await session.exec(select(UpstreamProviderRow))
         providers_by_id = {p.id: p for p in provider_result.all()}
 
-        overrides_by_id: dict[str, tuple[ModelRow, float]] = {
-            row.id: (
+        overrides_by_key: dict[tuple[str, int], tuple[ModelRow, float]] = {
+            (row.id.lower(), row.upstream_provider_id): (
                 row,
-                providers_by_id[row.upstream_provider_id].provider_fee
-                if row.upstream_provider_id in providers_by_id
-                else 1.01,
+                providers_by_id[row.upstream_provider_id].provider_fee,
             )
             for row in override_rows
             if row.upstream_provider_id is not None
@@ -106,17 +105,28 @@ async def get_all_models_with_overrides(
             and providers_by_id[row.upstream_provider_id].enabled
         }
 
-    all_models: dict[str, Model] = {}
+    all_models: dict[tuple[str, str], Model] = {}
 
     for upstream in upstreams:
+        upstream_db_id = getattr(upstream, "db_id", None)
+        provider_key = (
+            f"db:{upstream_db_id}"
+            if isinstance(upstream_db_id, int)
+            else f"{getattr(upstream, 'provider_type', '')}|{getattr(upstream, 'base_url', '')}"
+        )
         for model in upstream.get_cached_models():
-            if model.id in overrides_by_id:
-                override_row, provider_fee = overrides_by_id[model.id]
-                all_models[model.id] = _row_to_model(
+            model_key = (
+                (model.id.lower(), upstream_db_id)
+                if isinstance(upstream_db_id, int)
+                else None
+            )
+            if model_key is not None and model_key in overrides_by_key:
+                override_row, provider_fee = overrides_by_key[model_key]
+                all_models[(model.id.lower(), provider_key)] = _row_to_model(
                     override_row, apply_provider_fee=True, provider_fee=provider_fee
                 )
             elif model.enabled:
-                all_models[model.id] = model
+                all_models[(model.id.lower(), provider_key)] = model
 
     return list(all_models.values())
 
@@ -215,9 +225,6 @@ async def init_upstreams() -> list[BaseUpstreamProvider]:
 
             provider = _instantiate_provider(provider_row)
             if provider:
-                # Keep provider DB id on runtime instance so model mapping can
-                # bind DB overrides to the correct upstream.
-                setattr(provider, "db_id", provider_row.id)
                 await provider.refresh_models_cache()
                 logger.debug(
                     f"Initialized {provider_row.provider_type} provider",
@@ -250,6 +257,7 @@ async def _seed_providers_from_settings(
 
     providers_to_add: list[UpstreamProviderRow] = []
     seeded_provider_keys: set[tuple[str, str]] = set()
+    reserved_slugs: set[str] = set()
 
     provider_classes_by_type = {
         cls.provider_type: cls
@@ -264,6 +272,7 @@ async def _seed_providers_from_settings(
         ("PERPLEXITY_API_KEY", "perplexity", None, None),
         ("FIREWORKS_API_KEY", "fireworks", None, None),
         ("XAI_API_KEY", "xai", None, None),
+        ("TINFOIL_API_KEY", "tinfoil", None, None),
     ]
 
     for env_key, provider_type, _, _ in env_mappings:
@@ -279,8 +288,13 @@ async def _seed_providers_from_settings(
                     )
                 )
                 if not result.first():
+                    slug = await allocate_unique_provider_slug(
+                        session, provider_type, reserved_slugs
+                    )
+                    reserved_slugs.add(slug)
                     providers_to_add.append(
                         UpstreamProviderRow(
+                            slug=slug,
                             provider_type=provider_type,
                             base_url=base_url,
                             api_key=api_key,
@@ -299,8 +313,13 @@ async def _seed_providers_from_settings(
             )
         )
         if not result.first():
+            slug = await allocate_unique_provider_slug(
+                session, "ollama", reserved_slugs
+            )
+            reserved_slugs.add(slug)
             providers_to_add.append(
                 UpstreamProviderRow(
+                    slug=slug,
                     provider_type="ollama",
                     base_url=ollama_base_url,
                     api_key=ollama_api_key,
@@ -320,8 +339,13 @@ async def _seed_providers_from_settings(
                 )
             )
             if not result.first():
+                slug = await allocate_unique_provider_slug(
+                    session, "azure", reserved_slugs
+                )
+                reserved_slugs.add(slug)
                 providers_to_add.append(
                     UpstreamProviderRow(
+                        slug=slug,
                         provider_type="azure",
                         base_url=base_url,
                         api_key=api_key,
@@ -342,8 +366,13 @@ async def _seed_providers_from_settings(
                 )
             )
             if not result.first():
+                slug = await allocate_unique_provider_slug(
+                    session, "custom", reserved_slugs
+                )
+                reserved_slugs.add(slug)
                 providers_to_add.append(
                     UpstreamProviderRow(
+                        slug=slug,
                         provider_type="custom",
                         base_url=base_url,
                         api_key=api_key,
@@ -356,7 +385,7 @@ async def _seed_providers_from_settings(
         session.add(provider)
         logger.info(
             f"Seeding {provider.provider_type} provider",  # type: ignore[str-format]
-            extra={"base_url": provider.base_url},
+            extra={"base_url": provider.base_url, "slug": provider.slug},
         )
 
 
@@ -391,9 +420,7 @@ def _instantiate_provider(
             return provider
 
         if provider_row.provider_type == "custom":
-            return BaseUpstreamProvider(
-                provider_row.base_url, provider_row.api_key, provider_row.provider_fee
-            )
+            return BaseUpstreamProvider.from_db_row(provider_row)
 
         logger.error(
             f"Unknown provider type: {provider_row.provider_type}",

routstr/upstream/messages_dispatch.py

@@ -29,7 +29,9 @@ import litellm
 
 from ..core import get_logger
 from ..core.exceptions import UpstreamError
+from ..core.redaction import redact_org_ids
 from ..payment.models import Model
+from .rate_limit import classify_rate_limit
 
 logger = get_logger(__name__)
 
@@ -505,23 +507,33 @@ async def dispatch_anthropic_messages(
     try:
         result = await litellm.anthropic.messages.acreate(**kwargs)
     except Exception as exc:
-        exc_message = getattr(exc, "message", None) or str(exc) or repr(exc)
+        raw_message = getattr(exc, "message", None) or str(exc) or repr(exc)
+        # Redact provider account identifiers before the message reaches logs
+        # or the surfaced error.
+        exc_message = redact_org_ids(raw_message)
         exc_status = getattr(exc, "status_code", None)
         exc_response = getattr(exc, "response", None)
         response_text = None
         if exc_response is not None:
             try:
-                response_text = getattr(exc_response, "text", str(exc_response))
+                response_text = redact_org_ids(
+                    getattr(exc_response, "text", str(exc_response))
+                )
             except Exception:
                 response_text = "<unreadable>"
+        status_for_classify = exc_status if isinstance(exc_status, int) else 502
+        rate_limit = classify_rate_limit(
+            status_for_classify, exc_message, getattr(exc, "headers", None)
+        )
         logger.error(
             "litellm dispatch failed",
             extra={
                 "error": exc_message,
                 "error_type": type(exc).__name__,
                 "status_code": exc_status,
+                "error_code": rate_limit.code if rate_limit else None,
                 "llm_provider": getattr(exc, "llm_provider", None),
-                "body": getattr(exc, "body", None),
+                "body": redact_org_ids(str(getattr(exc, "body", "") or "")) or None,
                 "response_text": response_text,
                 "model": litellm_model,
                 "api_base": base_url,
@@ -529,7 +541,9 @@ async def dispatch_anthropic_messages(
         )
         raise UpstreamError(
             f"Upstream error via litellm: {exc_message}",
-            status_code=exc_status if isinstance(exc_status, int) else 502,
+            status_code=status_for_classify,
+            code=rate_limit.code if rate_limit else None,
+            details=rate_limit.as_details() if rate_limit else None,
         ) from exc
 
     if not client_stream and hasattr(result, "__aiter__"):

routstr/upstream/ollama.py

@@ -43,7 +43,7 @@ class OllamaUpstreamProvider(BaseUpstreamProvider):
         )
 
     @classmethod
-    def from_db_row(
+    def _build_from_row(
         cls, provider_row: "UpstreamProviderRow"
     ) -> "OllamaUpstreamProvider":
         return cls(

routstr/upstream/openai.py

@@ -20,7 +20,7 @@ class OpenAIUpstreamProvider(BaseUpstreamProvider):
         )
 
     @classmethod
-    def from_db_row(
+    def _build_from_row(
         cls, provider_row: "UpstreamProviderRow"
     ) -> "OpenAIUpstreamProvider":
         return cls(

routstr/upstream/openrouter.py

@@ -57,7 +57,7 @@ class OpenRouterUpstreamProvider(BaseUpstreamProvider):
         )
 
     @classmethod
-    def from_db_row(
+    def _build_from_row(
         cls, provider_row: "UpstreamProviderRow"
     ) -> "OpenRouterUpstreamProvider":
         return cls(

routstr/upstream/perplexity.py

@@ -23,7 +23,7 @@ class PerplexityUpstreamProvider(BaseUpstreamProvider):
         )
 
     @classmethod
-    def from_db_row(
+    def _build_from_row(
         cls, provider_row: "UpstreamProviderRow"
     ) -> "PerplexityUpstreamProvider":
         return cls(

routstr/upstream/ppqai.py

@@ -8,6 +8,7 @@ from pydantic.v1 import BaseModel, Field
 from ..core.logging import get_logger
 from ..payment.models import Architecture, Model, Pricing, async_fetch_openrouter_models
 from .base import BaseUpstreamProvider, TopupData
+from .ehbp import EHBPForwardingTarget
 
 if TYPE_CHECKING:
     from ..core.db import UpstreamProviderRow
@@ -39,6 +40,10 @@ class PPQAIUpstreamProvider(BaseUpstreamProvider):
     default_base_url = "https://api.ppq.ai"
     platform_url = "https://ppq.ai/api-docs"
     IGNORED_MODEL_IDS: list[str] = ["auto"]
+    # PPQ.AI has a private encrypted endpoint, but this proxy currently has no
+    # provider-attested usage extractor/model binding for it. Keep EHBP disabled
+    # until a ConfidentialInferenceProfile can bill it without max-cost fallback.
+    supports_ehbp = False
 
     def __init__(self, api_key: str, provider_fee: float = 1.0):
         super().__init__(
@@ -46,7 +51,7 @@ class PPQAIUpstreamProvider(BaseUpstreamProvider):
         )
 
     @classmethod
-    def from_db_row(
+    def _build_from_row(
         cls, provider_row: "UpstreamProviderRow"
     ) -> "PPQAIUpstreamProvider":
         return cls(
@@ -70,6 +75,20 @@ class PPQAIUpstreamProvider(BaseUpstreamProvider):
     def transform_model_name(self, model_id: str) -> str:
         return model_id
 
+    def get_ehbp_forwarding_target(
+        self, path: str, model_obj: Model
+    ) -> EHBPForwardingTarget:
+        """Return the PPQ.AI private enclave target for EHBP requests.
+
+        PPQ.AI exposes EHBP-aware inference under /private/v1/... separate
+        from the public /v1/... endpoint. The encrypted body remains opaque to
+        Routstr, so PPQ.AI also needs X-Private-Model for routing/billing.
+        """
+        return EHBPForwardingTarget(
+            url=f"{self.base_url.rstrip('/')}/private/{path.lstrip('/')}",
+            headers={"X-Private-Model": model_obj.forwarded_model_id or model_obj.id},
+        )
+
     @classmethod
     async def create_account_static(cls) -> dict[str, object]:
         """Create a new PPQ.AI account without requiring an instance.
@@ -229,17 +248,14 @@ class PPQAIUpstreamProvider(BaseUpstreamProvider):
                 f"Disabling PPQ.AI provider ({self.base_url}) due to insufficient balance",
                 extra={"error": error_message},
             )
-            from sqlmodel import select
-
             from ..core.db import UpstreamProviderRow, create_session
 
             async with create_session() as session:
-                statement = select(UpstreamProviderRow).where(
-                    UpstreamProviderRow.base_url == self.base_url,
-                    UpstreamProviderRow.api_key == self.api_key,
+                provider = (
+                    await session.get(UpstreamProviderRow, self.db_id)
+                    if self.db_id is not None
+                    else None
                 )
-                result = await session.exec(statement)
-                provider = result.first()
 
                 if provider:
                     provider.enabled = False

routstr/upstream/pricing_resolver.py

@@ -0,0 +1,203 @@
+"""Shared price/metadata resolution chain for upstream model discovery.
+
+Most OpenAI-compatible ``/models`` responses carry no pricing. Rather than let
+a provider fabricate one, this module resolves a model through decreasingly
+trustworthy sources — litellm's bundled cost map (curated list prices, mirrors
+provider docs), then the OpenRouter feed (resale prices, broader coverage) —
+and returns ``None`` when none of them know the model, so the caller can fail
+closed instead of inventing a number.
+
+Provider-native pricing (a gateway's own ``/models`` schema, e.g. Venice's
+``model_spec``) is authoritative and handled by the provider before this chain
+is consulted; only the shared fallback lives here so a later refactor can hoist
+it into the base provider unchanged.
+"""
+
+from __future__ import annotations
+
+from dataclasses import dataclass, field
+
+
+@dataclass
+class ResolvedPricing:
+    """Per-token pricing plus whatever metadata the answering source carried.
+
+    Prices are USD per token. ``source`` records provenance
+    (``native``/``litellm``/``openrouter``/``unresolved``) so later work can
+    surface where each price came from.
+    """
+
+    prompt: float
+    completion: float
+    context_length: int | None
+    source: str
+    modality: str | None = None
+    max_completion_tokens: int | None = None
+    input_cache_read: float = 0.0
+    input_cache_write: float = 0.0
+    input_modalities: list[str] = field(default_factory=lambda: ["text"])
+    output_modalities: list[str] = field(default_factory=lambda: ["text"])
+    tokenizer: str = "unknown"
+    instruct_type: str | None = None
+    is_moderated: bool | None = None
+
+
+def estimate_context_length(model_id: str) -> int:
+    """Best-effort context window from a model id when no source reports one.
+
+    The last rung of the fallback chain, reached only for a model whose price
+    resolved but whose context did not (or that imported disabled). Context is
+    not a billing input, so a rough id-based guess is acceptable here where a
+    guessed *price* never would be.
+    """
+    lowered = model_id.lower()
+    if any(pattern in lowered for pattern in ["32k", "32000"]):
+        return 32768
+    if any(pattern in lowered for pattern in ["16k", "16000"]):
+        return 16384
+    if any(pattern in lowered for pattern in ["8k", "8000"]):
+        return 8192
+    if "gpt-4" in lowered:
+        return 8192
+    if "claude" in lowered:
+        return 200000
+    return 4096
+
+
+def _as_float(value: object) -> float | None:
+    """OpenRouter reports prices as strings; coerce, ``None`` if unparseable."""
+    try:
+        return float(value)  # type: ignore[arg-type]
+    except (TypeError, ValueError):
+        return None
+
+
+def _as_int(value: object) -> int | None:
+    """Coerce an already-numeric token count to ``int``, else ``None``."""
+    return int(value) if isinstance(value, (int, float)) else None
+
+
+def _from_litellm(model_id: str) -> ResolvedPricing | None:
+    # Lazy import so the resolver stays import-light and shares the exact
+    # lookup semantics used by cache-rate backfill.
+    from ..payment.models import litellm_cost_entry
+
+    info = litellm_cost_entry(model_id)
+    if info is None:
+        return None
+
+    prompt = info.get("input_cost_per_token")
+    completion = info.get("output_cost_per_token")
+    if not isinstance(prompt, (int, float)) or not isinstance(completion, (int, float)):
+        return None
+    # A both-zero entry is litellm listing a model without a real price (free
+    # moderation/rerank tiers do this) — treating 0/0 as resolved would serve
+    # the model for free. Reject it (and any negative) so the caller falls
+    # through, mirroring async_fetch_openrouter_models' _has_valid_pricing.
+    if prompt < 0 or completion < 0 or (prompt == 0 and completion == 0):
+        return None
+
+    input_modalities = ["text"]
+    if info.get("supports_vision"):
+        input_modalities.append("image")
+
+    return ResolvedPricing(
+        prompt=float(prompt),
+        completion=float(completion),
+        # max_input_tokens is the context window; max_tokens is litellm's
+        # completion cap (it tracks max_output_tokens for ~94% of models), so
+        # it is never a context source. A missing window falls to the id-based
+        # estimate downstream rather than borrowing the output cap.
+        context_length=_as_int(info.get("max_input_tokens")),
+        source="litellm",
+        max_completion_tokens=_as_int(info.get("max_output_tokens")),
+        input_cache_read=float(info.get("cache_read_input_token_cost") or 0.0),
+        input_cache_write=float(info.get("cache_creation_input_token_cost") or 0.0),
+        input_modalities=input_modalities,
+    )
+
+
+def _match_openrouter(model_id: str, feed: list[dict]) -> dict | None:
+    """Find ``model_id`` in the OpenRouter feed, exact id before bare tail.
+
+    Bare-tail matching (``deepseek-chat`` ↔ ``deepseek/deepseek-chat``) is a
+    looser, lower-trust match — OpenRouter fans a model out across resellers —
+    so an exact id match always wins first. When several entries share the bare
+    tail, the one with the highest *combined* (prompt + completion) per-token
+    cost wins: the choice must be deterministic (not feed-order-dependent) and
+    money-safe whichever way traffic leans, since undercharging is the hazard.
+    Ranking on prompt alone could pick an entry that is cheap on input but dear
+    on output. The live feed has no such collisions today; this only governs
+    the latent case.
+    """
+    bare = model_id.split("/", 1)[-1]
+    exact = next((m for m in feed if m.get("id") == model_id), None)
+    if exact is not None:
+        return exact
+    matches = [m for m in feed if m.get("id", "").split("/", 1)[-1] == bare]
+    if not matches:
+        return None
+
+    def _combined_cost(m: dict) -> float:
+        pricing = m.get("pricing", {})
+        return (_as_float(pricing.get("prompt")) or 0.0) + (
+            _as_float(pricing.get("completion")) or 0.0
+        )
+
+    return max(matches, key=_combined_cost)
+
+
+def _from_openrouter(model_id: str, feed: list[dict]) -> ResolvedPricing | None:
+    entry = _match_openrouter(model_id, feed)
+    if entry is None:
+        return None
+
+    pricing = entry.get("pricing", {})
+    prompt = _as_float(pricing.get("prompt"))
+    completion = _as_float(pricing.get("completion"))
+    if prompt is None or completion is None:
+        return None
+
+    architecture = entry.get("architecture", {})
+    top_provider = entry.get("top_provider", {})
+
+    return ResolvedPricing(
+        prompt=prompt,
+        completion=completion,
+        context_length=_as_int(entry.get("context_length")),
+        source="openrouter",
+        modality=architecture.get("modality"),
+        max_completion_tokens=_as_int(top_provider.get("max_completion_tokens")),
+        input_cache_read=_as_float(pricing.get("input_cache_read")) or 0.0,
+        input_cache_write=_as_float(pricing.get("input_cache_write")) or 0.0,
+        input_modalities=architecture.get("input_modalities") or ["text"],
+        output_modalities=architecture.get("output_modalities") or ["text"],
+        tokenizer=architecture.get("tokenizer") or "unknown",
+        instruct_type=architecture.get("instruct_type"),
+        is_moderated=top_provider.get("is_moderated"),
+    )
+
+
+class FallbackPricingResolver:
+    """Resolves models via litellm → OpenRouter for one discovery pass.
+
+    The OpenRouter catalog is fetched at most once and only when a model
+    actually misses litellm, so a provider full of litellm-known models never
+    touches the network. Instantiate one per ``fetch_models`` call.
+    """
+
+    def __init__(self) -> None:
+        self._openrouter_feed: list[dict] | None = None
+
+    async def resolve(self, model_id: str) -> ResolvedPricing | None:
+        """Resolve ``model_id``; ``None`` if no source knows it."""
+        resolved = _from_litellm(model_id)
+        if resolved is not None:
+            return resolved
+
+        if self._openrouter_feed is None:
+            # Lazy import so tests can patch the feed at its source.
+            from ..payment.models import async_fetch_openrouter_models
+
+            self._openrouter_feed = await async_fetch_openrouter_models()
+        return _from_openrouter(model_id, self._openrouter_feed)

routstr/upstream/rate_limit.py

@@ -0,0 +1,136 @@
+"""Detection and parsing of upstream provider rate-limit errors.
+
+Upstream OpenAI-compatible providers signal rate limits via HTTP 429 and/or a
+human-readable message such as::
+
+    Rate limit reached for gpt-5.5-2026-04-23 (for limit gpt-5.5) in organization
+    org-XXXX on tokens per min (TPM): Limit 180000000, Used 180000000,
+    Requested 8929. Please try again in 2ms.
+
+This module classifies those failures into a stable :data:`UPSTREAM_RATE_LIMIT`
+code and extracts useful debugging fields. All retained text is redacted of
+organization IDs first.
+"""
+
+from __future__ import annotations
+
+import re
+from dataclasses import asdict, dataclass
+
+from ..core.redaction import redact_org_ids
+
+# Stable error code callers can switch on to distinguish upstream rate limits
+# from generic request failures. The literal value matches the identifier named
+# in issue #555 ("UPSTREAM_RATE_LIMIT") so the public API contract is exact.
+UPSTREAM_RATE_LIMIT = "UPSTREAM_RATE_LIMIT"
+
+# Message fragments that indicate a rate-limit even when the status code is not
+# 429 (some providers wrap it in a 400/500 envelope).
+_RATE_LIMIT_MARKERS = (
+    "rate limit reached",
+    "rate_limit_exceeded",
+    "rate limit exceeded",
+    "too many requests",
+)
+
+_MODEL_RE = re.compile(r"Rate limit reached for ([^\s(]+)", re.IGNORECASE)
+_LIMIT_NAME_RE = re.compile(r"\(for limit ([^)]+)\)", re.IGNORECASE)
+_METRIC_RE = re.compile(r"on ([a-z ]+\((?:TPM|RPM|TPD|RPD|IPM)\))", re.IGNORECASE)
+_LIMIT_RE = re.compile(r"Limit (\d+)", re.IGNORECASE)
+_USED_RE = re.compile(r"Used (\d+)", re.IGNORECASE)
+_REQUESTED_RE = re.compile(r"Requested (\d+)", re.IGNORECASE)
+_RETRY_RE = re.compile(r"try again in ([\d.]+)\s*(ms|s)", re.IGNORECASE)
+
+
+@dataclass
+class RateLimitInfo:
+    """Structured, redaction-safe view of an upstream rate-limit error."""
+
+    code: str
+    message: str
+    model: str | None = None
+    limit_name: str | None = None
+    metric: str | None = None
+    limit: int | None = None
+    used: int | None = None
+    requested: int | None = None
+    retry_after_seconds: float | None = None
+
+    def as_details(self) -> dict[str, object]:
+        """Return a JSON-serialisable dict for embedding in an error envelope."""
+        return {k: v for k, v in asdict(self).items() if v is not None}
+
+
+def _looks_like_rate_limit(status_code: int, message: str) -> bool:
+    if status_code == 429:
+        return True
+    lowered = message.lower()
+    return any(marker in lowered for marker in _RATE_LIMIT_MARKERS)
+
+
+def _parse_retry_after_header(headers: dict[str, str] | None) -> float | None:
+    """Parse a ``Retry-After`` header (delta-seconds form) into seconds."""
+    if not headers:
+        return None
+    raw = headers.get("retry-after") or headers.get("Retry-After")
+    if raw is None:
+        return None
+    try:
+        return float(str(raw).strip())
+    except (TypeError, ValueError):
+        return None
+
+
+def _int_or_none(match: re.Match[str] | None) -> int | None:
+    if match is None:
+        return None
+    try:
+        return int(match.group(1))
+    except (TypeError, ValueError):
+        return None
+
+
+def classify_rate_limit(
+    status_code: int,
+    message: str,
+    headers: dict[str, str] | None = None,
+) -> RateLimitInfo | None:
+    """Classify an upstream error as a rate-limit and extract its fields.
+
+    Args:
+        status_code: HTTP status code from the upstream response.
+        message: Upstream error message (may contain sensitive identifiers).
+        headers: Optional upstream response headers, used for ``Retry-After``.
+
+    Returns:
+        A :class:`RateLimitInfo` when the error is a rate-limit, else ``None``.
+    """
+    message = message or ""
+    if not _looks_like_rate_limit(status_code, message):
+        return None
+
+    redacted = redact_org_ids(message)
+
+    model_match = _MODEL_RE.search(redacted)
+    metric_match = _METRIC_RE.search(redacted)
+
+    retry_after = _parse_retry_after_header(headers)
+    if retry_after is None:
+        retry_match = _RETRY_RE.search(redacted)
+        if retry_match is not None:
+            value = float(retry_match.group(1))
+            retry_after = value / 1000.0 if retry_match.group(2).lower() == "ms" else value
+
+    limit_name_match = _LIMIT_NAME_RE.search(redacted)
+
+    return RateLimitInfo(
+        code=UPSTREAM_RATE_LIMIT,
+        message=redacted,
+        model=model_match.group(1) if model_match else None,
+        limit_name=limit_name_match.group(1).strip() if limit_name_match else None,
+        metric=metric_match.group(1).strip() if metric_match else None,
+        limit=_int_or_none(_LIMIT_RE.search(redacted)),
+        used=_int_or_none(_USED_RE.search(redacted)),
+        requested=_int_or_none(_REQUESTED_RE.search(redacted)),
+        retry_after_seconds=retry_after,
+    )

routstr/upstream/routstr.py

@@ -55,7 +55,7 @@ class RoutstrUpstreamProvider(BaseUpstreamProvider):
         return path.lstrip("/")
 
     @classmethod
-    def from_db_row(
+    def _build_from_row(
         cls, provider_row: "UpstreamProviderRow"
     ) -> "RoutstrUpstreamProvider":
         import json

routstr/upstream/tinfoil.py

@@ -0,0 +1,236 @@
+from __future__ import annotations
+
+from typing import TYPE_CHECKING
+
+import httpx
+from fastapi import Request
+from fastapi.responses import Response, StreamingResponse
+from pydantic.v1 import BaseModel
+
+from ..core.exceptions import UpstreamError
+from ..core.logging import get_logger
+from ..payment.models import Architecture, Model, Pricing
+from .base import BaseUpstreamProvider
+from .ehbp import (
+    _ENCLAVE_URL_HEADER,
+    _PROXY_ONLY_HEADERS,
+    _RESPONSE_USAGE_HEADER,
+    ConfidentialInferenceProfile,
+    EHBPForwardingTarget,
+)
+
+if TYPE_CHECKING:
+    from ..core.db import UpstreamProviderRow
+
+logger = get_logger(__name__)
+
+
+class TinfoilModelPricing(BaseModel):
+    inputTokenPricePer1M: float = 0.0
+    outputTokenPricePer1M: float = 0.0
+    requestPrice: float = 0.0
+
+
+class TinfoilModel(BaseModel):
+    id: str
+    context_window: int = 0
+    created: int = 0
+    multimodal: bool = False
+    reasoning: bool = False
+    tool_calling: bool = False
+    type: str = "chat"
+    pricing: TinfoilModelPricing = TinfoilModelPricing()
+    endpoints: list[str] = []
+
+
+class TinfoilUpstreamProvider(BaseUpstreamProvider):
+    """Direct upstream provider for the Tinfoil inference API.
+
+    Tinfoil hosts open-source models inside attested secure enclaves and exposes
+    an OpenAI-compatible API at ``https://inference.tinfoil.sh``. Request and
+    response bodies are encrypted end-to-end with EHBP (HPKE), so Routstr acts
+    as a blind relay: it forwards the opaque encrypted body, never sees
+    plaintext, and bills from the ``X-Tinfoil-Usage-Metrics`` header that
+    Tinfoil returns outside the encrypted body when
+    ``X-Tinfoil-Request-Usage-Metrics: true`` is set.
+    """
+
+    provider_type = "tinfoil"
+    default_base_url = "https://inference.tinfoil.sh"
+    platform_url = "https://docs.tinfoil.sh"
+    supports_ehbp = True
+    confidential_inference_profile = ConfidentialInferenceProfile(
+        usage_response_header=_RESPONSE_USAGE_HEADER,
+        client_target_url_header=_ENCLAVE_URL_HEADER,
+        allow_client_target_override=True,
+        proxy_only_headers=_PROXY_ONLY_HEADERS,
+    )
+
+    def __init__(self, api_key: str, provider_fee: float = 1.0):
+        super().__init__(
+            base_url=self.default_base_url,
+            api_key=api_key,
+            provider_fee=provider_fee,
+        )
+
+    @classmethod
+    def _build_from_row(
+        cls, provider_row: "UpstreamProviderRow"
+    ) -> "TinfoilUpstreamProvider":
+        return cls(
+            api_key=provider_row.api_key,
+            provider_fee=provider_row.provider_fee,
+        )
+
+    @classmethod
+    def get_provider_metadata(cls) -> dict[str, object]:
+        return {
+            "id": cls.provider_type,
+            "name": "Tinfoil",
+            "default_base_url": cls.default_base_url,
+            "fixed_base_url": True,
+            "platform_url": cls.platform_url,
+            "can_create_account": False,
+            "can_topup": False,
+            "can_show_balance": False,
+        }
+
+    def transform_model_name(self, model_id: str) -> str:
+        return model_id.removeprefix("tinfoil/")
+
+    def get_confidential_inference_profile(self) -> ConfidentialInferenceProfile:
+        return self.confidential_inference_profile
+
+    async def forward_get_request(
+        self,
+        request: Request,
+        path: str,
+        headers: dict,
+    ) -> Response | StreamingResponse:
+        """Handle Tinfoil-specific GET endpoints.
+
+        * ``/attestation`` (or ``/tee/attestation``): proxy to the Tinfoil ATC
+          (attestation bundle proxy) at ``https://atc.tinfoil.sh/attestation``.
+        * Other GETs: forward to the provider base URL
+          (``https://inference.tinfoil.sh``). ``X-Tinfoil-Enclave-Url`` is an
+          EHBP-only header used for encrypted POST requests and is not honored
+          for unencrypted GET requests.
+        """
+        clean_path = path.removeprefix("tee/").rstrip("/")
+        if clean_path == "attestation":
+            return await self._proxy_attestation(headers)
+        return await super().forward_get_request(request, path, headers)
+
+    async def _proxy_attestation(self, headers: dict) -> Response:
+        url = "https://atc.tinfoil.sh/attestation"
+        async with httpx.AsyncClient(
+            transport=httpx.AsyncHTTPTransport(retries=1),
+            timeout=30.0,
+        ) as client:
+            try:
+                resp = await client.get(
+                    url,
+                    headers={
+                        "Accept": headers.get("accept", "application/json"),
+                    },
+                )
+                response_headers = dict(resp.headers)
+                response_headers.pop("content-encoding", None)
+                response_headers.pop("content-length", None)
+                return Response(
+                    content=resp.content,
+                    status_code=resp.status_code,
+                    headers=response_headers,
+                )
+            except Exception as exc:
+                raise UpstreamError(
+                    f"Error fetching Tinfoil attestation: {type(exc).__name__}",
+                    status_code=502,
+                ) from exc
+
+    def get_ehbp_forwarding_target(
+        self, path: str, model_obj: Model
+    ) -> EHBPForwardingTarget:
+        """Return the Tinfoil enclave target for EHBP requests.
+
+        Requests usage metrics from the enclave so Routstr can bill exactly
+        without decrypting the response body. The actual forwarding URL is
+        overridden at dispatch time by ``X-Tinfoil-Enclave-Url`` when the SDK
+        sends it (see ``routstr/upstream/ehbp.py``).
+        """
+        return EHBPForwardingTarget(
+            url=f"{self.base_url.rstrip('/')}/{path.lstrip('/')}",
+            headers={"X-Tinfoil-Request-Usage-Metrics": "true"},
+            profile=self.confidential_inference_profile,
+        )
+
+    async def fetch_models(self) -> list[Model]:
+        """Fetch models from the public Tinfoil models endpoint.
+
+        ``GET /v1/models`` is unauthenticated and returns all available models
+        with their pricing in USD per 1M tokens.
+        """
+        url = f"{self.base_url}/v1/models"
+        try:
+            async with httpx.AsyncClient(timeout=30.0) as client:
+                response = await client.get(url)
+                response.raise_for_status()
+                data = response.json()
+                models_data = data.get("data", [])
+
+                models: list[Model] = []
+                for model_data in models_data:
+                    try:
+                        tf = TinfoilModel.parse_obj(model_data)
+                        input_price = tf.pricing.inputTokenPricePer1M
+                        output_price = tf.pricing.outputTokenPricePer1M
+                        request_price = tf.pricing.requestPrice
+
+                        modality = "text->text"
+                        input_modalities = ["text"]
+                        output_modalities = ["text"]
+                        if tf.multimodal:
+                            modality = "text->text+image"
+                            input_modalities = ["text", "image"]
+
+                        models.append(
+                            Model(
+                                id=tf.id,
+                                name=tf.id,
+                                created=tf.created,
+                                description=f"Tinfoil {tf.type} model",
+                                context_length=tf.context_window,
+                                architecture=Architecture(
+                                    modality=modality,
+                                    input_modalities=input_modalities,
+                                    output_modalities=output_modalities,
+                                    tokenizer="Unknown",
+                                    instruct_type=None,
+                                ),
+                                pricing=Pricing(
+                                    prompt=input_price / 1_000_000,
+                                    completion=output_price / 1_000_000,
+                                    request=request_price,
+                                    image=0.0,
+                                    web_search=0.0,
+                                    internal_reasoning=0.0,
+                                ),
+                            )
+                        )
+                    except Exception as e:
+                        logger.warning(
+                            "Failed to parse Tinfoil model",
+                            extra={
+                                "model_id": model_data.get("id", "unknown"),
+                                "error": str(e),
+                                "error_type": type(e).__name__,
+                            },
+                        )
+
+                return models
+        except Exception as e:
+            logger.error(
+                "Error fetching models from Tinfoil",
+                extra={"error": str(e), "error_type": type(e).__name__},
+            )
+            return []

routstr/upstream/tinfoil_trailer.py

@@ -0,0 +1,189 @@
+"""h11-based HTTP client for EHBP requests that captures HTTP trailers.
+
+httpx/httpcore silently discard HTTP trailers during chunked transfer
+decoding. Tinfoil returns ``X-Tinfoil-Usage-Metrics`` as a trailer on
+streaming responses, so we need a lower-level HTTP client that preserves
+trailers from the h11 ``EndOfMessage`` event.
+
+Because EHBP response bodies are opaque encrypted blobs, buffering the full
+response is acceptable — the client decrypts the complete body regardless of
+whether it arrived streamed or buffered.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import ssl
+from dataclasses import dataclass, field
+from urllib.parse import urlsplit
+
+import h11
+
+from ..core import get_logger
+
+logger = get_logger(__name__)
+
+_READ_BUFSIZE = 65536
+_DEFAULT_TIMEOUT_SECONDS = 30.0
+_DEFAULT_CLOSE_TIMEOUT_SECONDS = 1.0
+_DEFAULT_MAX_RESPONSE_BYTES = 25 * 1024 * 1024
+_HOP_BY_HOP_HEADERS = {
+    "connection",
+    "keep-alive",
+    "proxy-authenticate",
+    "proxy-authorization",
+    "te",
+    "trailer",
+    "transfer-encoding",
+    "upgrade",
+}
+
+
+@dataclass
+class TrailerResponse:
+    """Buffered HTTP response with optional trailer headers."""
+
+    status_code: int
+    headers: list[tuple[str, str]]
+    body: bytes
+    trailers: list[tuple[str, str]] = field(default_factory=list)
+
+
+def _get_header(headers: list[tuple[str, str]], name: str) -> str | None:
+    name_lower = name.lower()
+    for k, v in headers:
+        if k.lower() == name_lower:
+            return v
+    return None
+
+
+def _strip_hop_by_hop_headers(headers: dict[str, str]) -> dict[str, str]:
+    """Remove connection-specific headers before serializing a new request."""
+    connection_tokens: set[str] = set()
+    for key, value in headers.items():
+        if key.lower() == "connection":
+            connection_tokens.update(
+                token.strip().lower() for token in value.split(",") if token.strip()
+            )
+
+    excluded = _HOP_BY_HOP_HEADERS | connection_tokens
+    return {key: value for key, value in headers.items() if key.lower() not in excluded}
+
+
+async def forward_with_trailer(
+    *,
+    method: str,
+    url: str,
+    headers: dict[str, str],
+    body: bytes,
+    timeout_seconds: float = _DEFAULT_TIMEOUT_SECONDS,
+    max_response_bytes: int = _DEFAULT_MAX_RESPONSE_BYTES,
+    close_timeout_seconds: float = _DEFAULT_CLOSE_TIMEOUT_SECONDS,
+) -> TrailerResponse:
+    """Send an HTTP/1.1 request via h11 and capture HTTP trailers.
+
+    Returns a :class:`TrailerResponse` with the full buffered body and any
+    trailer headers from the ``EndOfMessage`` event.
+    """
+    parsed = urlsplit(url)
+    host = parsed.hostname
+    if not host:
+        raise ValueError(f"Invalid URL (no hostname): {url}")
+    port = parsed.port or 443
+    path = parsed.path or "/"
+    if parsed.query:
+        path = f"{path}?{parsed.query}"
+
+    # FastAPI has already decoded the incoming request body. Do not carry the
+    # original connection's framing or other hop-by-hop metadata into the new
+    # upstream connection.
+    headers = _strip_hop_by_hop_headers(headers)
+
+    ssl_ctx = ssl.create_default_context()
+    reader, writer = await asyncio.wait_for(
+        asyncio.open_connection(host, port, ssl=ssl_ctx),
+        timeout=timeout_seconds,
+    )
+
+    try:
+        # Build HTTP/1.1 request
+        header_lines = [f"{method} {path} HTTP/1.1"]
+        has_host = any(k.lower() == "host" for k in headers)
+        if not has_host:
+            header_lines.append(f"Host: {host}")
+        header_lines.append("Connection: close")
+
+        for key, value in headers.items():
+            if key.lower() == "host":
+                continue
+            header_lines.append(f"{key}: {value}")
+
+        if body and not any(k.lower() == "content-length" for k in headers):
+            header_lines.append(f"Content-Length: {len(body)}")
+
+        request_data = "\r\n".join(header_lines).encode() + b"\r\n\r\n"
+        if body:
+            request_data += body
+
+        writer.write(request_data)
+        await asyncio.wait_for(writer.drain(), timeout=timeout_seconds)
+
+        # Parse response with h11
+        conn = h11.Connection(h11.CLIENT)
+        status_code = 0
+        resp_headers: list[tuple[str, str]] = []
+        body_chunks: list[bytes] = []
+        body_size = 0
+        trailers: list[tuple[str, str]] = []
+
+        while True:
+            event = conn.next_event()
+
+            if event is h11.NEED_DATA:
+                data = await asyncio.wait_for(
+                    reader.read(_READ_BUFSIZE),
+                    timeout=timeout_seconds,
+                )
+                conn.receive_data(data if data else b"")
+                continue
+
+            if isinstance(event, h11.Response):
+                status_code = event.status_code
+                resp_headers = [(k.decode(), v.decode()) for k, v in event.headers]
+
+            elif isinstance(event, h11.Data):
+                body_size += len(event.data)
+                if body_size > max_response_bytes:
+                    raise ValueError(
+                        f"EHBP response exceeded {max_response_bytes} bytes"
+                    )
+                body_chunks.append(event.data)
+
+            elif isinstance(event, h11.EndOfMessage):
+                for k, v in event.headers:
+                    trailers.append((k.decode(), v.decode()))
+                break
+
+            elif isinstance(event, h11.PAUSED):
+                # Shouldn't happen for simple request/response, but break safely
+                logger.warning("h11 PAUSED event during EHBP response parsing")
+                break
+
+            elif isinstance(event, h11.ConnectionClosed):
+                break
+
+        return TrailerResponse(
+            status_code=status_code,
+            headers=resp_headers,
+            body=b"".join(body_chunks),
+            trailers=trailers,
+        )
+    finally:
+        writer.close()
+        if close_timeout_seconds > 0:
+            try:
+                await asyncio.wait_for(
+                    writer.wait_closed(), timeout=close_timeout_seconds
+                )
+            except Exception:
+                pass

routstr/upstream/xai.py

@@ -21,7 +21,7 @@ class XAIUpstreamProvider(BaseUpstreamProvider):
         )
 
     @classmethod
-    def from_db_row(cls, provider_row: "UpstreamProviderRow") -> "XAIUpstreamProvider":
+    def _build_from_row(cls, provider_row: "UpstreamProviderRow") -> "XAIUpstreamProvider":
         return cls(
             api_key=provider_row.api_key,
             provider_fee=provider_row.provider_fee,

scripts/__init__.py

@@ -0,0 +1 @@
+"""Operational and recovery scripts for routstr-core (not a shipped package)."""

scripts/reset_admin_password.py

@@ -0,0 +1,104 @@
+"""Recover admin access by resetting the stored admin password (issue #553).
+
+The lockout escape hatch for an operator who has lost the admin password. It
+talks to the ``secrets`` table directly and deliberately does *not* require
+``ROUTSTR_SECRET_KEY``: the admin password is scrypt-hashed (key-independent),
+so recovery works even when the encryption key is missing or has changed.
+
+Two explicit, mutually exclusive actions — running with no arguments only prints
+help, so the password can't be reset by accident:
+
+    python scripts/reset_admin_password.py --password <new-password>
+        Hash and store <new-password> now.
+
+    python scripts/reset_admin_password.py --regenerate
+        Clear the stored hash; the next node startup generates a fresh random
+        password and logs it once (with the /admin URL).
+"""
+
+import argparse
+import asyncio
+import sys
+import time
+
+from sqlmodel.ext.asyncio.session import AsyncSession
+
+from routstr.core.db import create_session, get_secret, set_admin_password
+from routstr.core.vault import MIN_PASSWORD_LENGTH
+
+
+def build_parser() -> argparse.ArgumentParser:
+    parser = argparse.ArgumentParser(
+        prog="reset_admin_password",
+        description="Reset the node's admin password (recovery from lockout).",
+    )
+    action = parser.add_mutually_exclusive_group()
+    action.add_argument(
+        "--password",
+        metavar="NEW_PASSWORD",
+        help=f"set this as the new admin password (min {MIN_PASSWORD_LENGTH} chars)",
+    )
+    action.add_argument(
+        "--regenerate",
+        action="store_true",
+        help="clear the password so the next startup generates and logs a new one",
+    )
+    return parser
+
+
+async def apply_reset(
+    session: AsyncSession,
+    *,
+    password: str | None = None,
+    regenerate: bool = False,
+) -> str:
+    """Perform the requested reset against ``session``; return a status message."""
+    if password is not None:
+        if len(password) < MIN_PASSWORD_LENGTH:
+            raise ValueError(
+                f"New password must be at least {MIN_PASSWORD_LENGTH} characters"
+            )
+        await set_admin_password(session, password)
+        return "Admin password updated."
+
+    if regenerate:
+        secret = await get_secret(session)
+        secret.admin_password_hash = None
+        secret.updated_at = int(time.time())
+        session.add(secret)
+        await session.commit()
+        return (
+            "Admin password cleared. The next node startup will generate a new "
+            "one and log it once with the /admin URL."
+        )
+
+    return ""
+
+
+async def _run(password: str | None, regenerate: bool) -> str:
+    async with create_session() as session:
+        return await apply_reset(
+            session, password=password, regenerate=regenerate
+        )
+
+
+def main(argv: list[str] | None = None) -> int:
+    parser = build_parser()
+    args = parser.parse_args(argv)
+
+    if args.password is None and not args.regenerate:
+        parser.print_help()
+        return 0
+
+    try:
+        message = asyncio.run(_run(args.password, args.regenerate))
+    except ValueError as exc:
+        print(f"error: {exc}", file=sys.stderr)
+        return 2
+
+    print(message)
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

tests/conftest.py

@@ -0,0 +1,15 @@
+"""Shared pytest configuration for the whole suite.
+
+A fixed, valid ``ROUTSTR_SECRET_KEY`` is set before any app import so that
+secret encryption is deterministic across the suite and the mandatory-key
+fail-fast does not break app-boot tests. Tests that need a different key (or an
+absent one) override this per-test via ``monkeypatch``.
+"""
+
+import os
+
+# Valid Fernet keys; KEY_A is the suite default, KEY_B is for wrong-key tests.
+TEST_SECRET_KEY = "l_Tkp-7xmjcQ-IFhr6qhILrU8HPRbEmYMrfSbo_5srU="
+TEST_SECRET_KEY_ALT = "_Teyrky_iToeDK51Tj1FsI9MJ340_cqKGmeher-a7MQ="
+
+os.environ.setdefault("ROUTSTR_SECRET_KEY", TEST_SECRET_KEY)

tests/integration/test_admin_auth.py

@@ -0,0 +1,121 @@
+"""Tests for admin password auth backed by the hashed Secret store (issue #553).
+
+Login and password change verify against the one-way ``Secret.admin_password_hash``
+(scrypt) instead of a plaintext settings field, which also closes the old ``!=``
+timing-attack comparison. The first hash is written by ``bootstrap_secrets`` at
+startup (generated, or migrated from a legacy password); here it is seeded
+directly into the store, then login checks against it and a password change
+re-hashes so the old password stops working and the new one starts.
+"""
+
+from __future__ import annotations
+
+import pytest
+from httpx import AsyncClient, Response
+
+from routstr.core.db import create_session, set_admin_password
+
+
+async def _seed_password(password: str) -> None:
+    # bootstrap_secrets owns first-password creation at startup; tests seed the
+    # hash straight into the store the same way, then exercise login/change.
+    async with create_session() as session:
+        await set_admin_password(session, password)
+
+
+async def _login(client: AsyncClient, password: str) -> Response:
+    return await client.post("/admin/api/login", json={"password": password})
+
+
+# --- login -----------------------------------------------------------------
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_login_500_when_no_password_configured(
+    integration_client: AsyncClient,
+) -> None:
+    resp = await _login(integration_client, "anything")
+    assert resp.status_code == 500
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_login_succeeds_with_correct_password(
+    integration_client: AsyncClient,
+) -> None:
+    await _seed_password("correct horse")
+    resp = await _login(integration_client, "correct horse")
+    assert resp.status_code == 200
+    body = resp.json()
+    assert body["ok"] is True
+    assert isinstance(body["token"], str) and body["token"]
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_login_rejects_wrong_password(
+    integration_client: AsyncClient,
+) -> None:
+    await _seed_password("correct horse")
+    resp = await _login(integration_client, "wrong horse")
+    assert resp.status_code == 401
+
+
+# --- password change -------------------------------------------------------
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_update_password_rehashes_so_only_new_works(
+    integration_client: AsyncClient,
+) -> None:
+    await _seed_password("old password")
+    login = await _login(integration_client, "old password")
+    token = login.json()["token"]
+    integration_client.headers["Authorization"] = f"Bearer {token}"
+
+    resp = await integration_client.patch(
+        "/admin/api/password",
+        json={"current_password": "old password", "new_password": "new password"},
+    )
+    assert resp.status_code == 200, resp.text
+
+    # Drop admin auth so the login calls aren't treated as authenticated noise.
+    integration_client.headers.pop("Authorization", None)
+    assert (await _login(integration_client, "old password")).status_code == 401
+    assert (await _login(integration_client, "new password")).status_code == 200
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_update_password_rejects_wrong_current(
+    integration_client: AsyncClient,
+) -> None:
+    await _seed_password("old password")
+    login = await _login(integration_client, "old password")
+    token = login.json()["token"]
+    integration_client.headers["Authorization"] = f"Bearer {token}"
+
+    resp = await integration_client.patch(
+        "/admin/api/password",
+        json={"current_password": "not the password", "new_password": "new password"},
+    )
+    assert resp.status_code == 401
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_update_password_rejects_short_new(
+    integration_client: AsyncClient,
+) -> None:
+    await _seed_password("old password")
+    login = await _login(integration_client, "old password")
+    token = login.json()["token"]
+    integration_client.headers["Authorization"] = f"Bearer {token}"
+
+    resp = await integration_client.patch(
+        "/admin/api/password",
+        json={"current_password": "old password", "new_password": "x"},
+    )
+    assert resp.status_code == 400

tests/integration/test_admin_model_cache_pricing.py

@@ -0,0 +1,229 @@
+import json
+from datetime import datetime, timedelta, timezone
+from unittest.mock import patch
+
+import pytest
+from httpx import AsyncClient
+from sqlmodel.ext.asyncio.session import AsyncSession
+
+from routstr.core.admin import admin_sessions
+from routstr.core.db import ModelRow, UpstreamProviderRow
+from routstr.payment.cost_calculation import CostData, calculate_cost
+from routstr.proxy import get_model_instance, reinitialize_upstreams
+
+
+def _admin_headers() -> dict[str, str]:
+    token = "test-admin-cache-pricing-token"
+    admin_sessions[token] = int(
+        (datetime.now(timezone.utc) + timedelta(minutes=5)).timestamp()
+    )
+    return {"Authorization": f"Bearer {token}"}
+
+
+def _model_payload(
+    provider_id: int,
+    *,
+    cache_read: float,
+    cache_write: float,
+) -> dict[str, object]:
+    return {
+        "id": "custom-cache-model",
+        "name": "Custom Cache Model",
+        "description": "custom model with explicit cache pricing",
+        "created": 0,
+        "context_length": 128000,
+        "architecture": {
+            "modality": "text",
+            "input_modalities": ["text"],
+            "output_modalities": ["text"],
+            "tokenizer": "unknown",
+            "instruct_type": None,
+        },
+        "pricing": {
+            "prompt": 1.4e-7,
+            "completion": 2.8e-7,
+            "input_cache_read": cache_read,
+            "input_cache_write": cache_write,
+            "request": 0.0,
+            "image": 0.0,
+            "web_search": 0.0,
+            "internal_reasoning": 0.0,
+        },
+        "per_request_limits": None,
+        "top_provider": None,
+        "upstream_provider_id": provider_id,
+        "canonical_slug": None,
+        "alias_ids": [],
+        "enabled": True,
+        "forwarded_model_id": "custom-cache-model",
+    }
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_admin_provider_model_api_persists_cache_pricing_on_create_and_update(
+    integration_client: AsyncClient,
+    integration_session: AsyncSession,
+) -> None:
+    provider = UpstreamProviderRow(
+        provider_type="generic",
+        base_url="https://custom-upstream.example/v1",
+        api_key="test-key",
+        provider_fee=1.0,
+    )
+    integration_session.add(provider)
+    await integration_session.commit()
+    await integration_session.refresh(provider)
+    assert provider.id is not None
+    await reinitialize_upstreams()
+
+    headers = _admin_headers()
+    create_payload = _model_payload(
+        provider.id,
+        cache_read=2.8e-9,
+        cache_write=3.5e-9,
+    )
+
+    with patch("routstr.payment.models.sats_usd_price", return_value=1e-6):
+        create_response = await integration_client.post(
+            f"/admin/api/upstream-providers/{provider.id}/models",
+            headers=headers,
+            json=create_payload,
+        )
+
+    assert create_response.status_code == 200
+    create_body = create_response.json()
+    assert create_body["pricing"]["input_cache_read"] == pytest.approx(2.8e-9)
+    assert create_body["pricing"]["input_cache_write"] == pytest.approx(3.5e-9)
+
+    row = await integration_session.get(ModelRow, ("custom-cache-model", provider.id))
+    assert row is not None
+    stored_pricing = json.loads(row.pricing)
+    assert stored_pricing["input_cache_read"] == pytest.approx(2.8e-9)
+    assert stored_pricing["input_cache_write"] == pytest.approx(3.5e-9)
+
+    update_payload = _model_payload(
+        provider.id,
+        cache_read=1.25e-9,
+        cache_write=4.5e-9,
+    )
+
+    with patch("routstr.payment.models.sats_usd_price", return_value=1e-6):
+        update_response = await integration_client.post(
+            f"/admin/api/upstream-providers/{provider.id}/models",
+            headers=headers,
+            json=update_payload,
+        )
+
+    assert update_response.status_code == 200
+    update_body = update_response.json()
+    assert update_body["pricing"]["input_cache_read"] == pytest.approx(1.25e-9)
+    assert update_body["pricing"]["input_cache_write"] == pytest.approx(4.5e-9)
+
+    await integration_session.refresh(row)
+    updated_pricing = json.loads(row.pricing)
+    assert updated_pricing["input_cache_read"] == pytest.approx(1.25e-9)
+    assert updated_pricing["input_cache_write"] == pytest.approx(4.5e-9)
+
+    model = get_model_instance("custom-cache-model")
+    assert model is not None
+    assert model.sats_pricing is not None
+    assert model.sats_pricing.input_cache_read == pytest.approx(0.00125)
+    assert model.sats_pricing.input_cache_write == pytest.approx(0.0045)
+
+    with patch("routstr.payment.cost_calculation.sats_usd_price", return_value=1e-6):
+        cost = await calculate_cost(
+            {
+                "model": "custom-cache-model",
+                "usage": {
+                    "prompt_tokens": 1000,
+                    "completion_tokens": 100,
+                    "prompt_tokens_details": {"cached_tokens": 800},
+                },
+            },
+            max_cost=1_000_000,
+        )
+
+    assert isinstance(cost, CostData)
+    assert cost.input_tokens == 200
+    assert cost.cache_read_input_tokens == 800
+    assert cost.cache_read_msats == 1000
+    assert cost.output_msats == 28000
+    assert cost.input_msats == 29000
+    assert cost.total_msats == 57000
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_upstream_response_cost_uses_model_cache_pricing(
+    integration_session: AsyncSession,
+    patched_db_engine: None,
+) -> None:
+    """A model's configured cache price must discount upstream cached-token usage."""
+    provider = UpstreamProviderRow(
+        provider_type="generic",
+        base_url="https://cache-priced-upstream.example/v1",
+        api_key="test-key",
+        provider_fee=1.0,
+    )
+    integration_session.add(provider)
+    await integration_session.commit()
+    await integration_session.refresh(provider)
+    assert provider.id is not None
+
+    row = ModelRow(
+        id="cache-priced-model",
+        name="Cache Priced Model",
+        description="model seeded with explicit cache pricing",
+        created=0,
+        context_length=128000,
+        architecture=json.dumps(
+            {
+                "modality": "text",
+                "input_modalities": ["text"],
+                "output_modalities": ["text"],
+                "tokenizer": "unknown",
+                "instruct_type": None,
+            }
+        ),
+        pricing=json.dumps(
+            {
+                "prompt": 1.4e-7,
+                "completion": 2.8e-7,
+                "input_cache_read": 1.25e-9,
+                "input_cache_write": 4.5e-9,
+            }
+        ),
+        upstream_provider_id=provider.id,
+        enabled=True,
+        forwarded_model_id="cache-priced-model",
+    )
+    integration_session.add(row)
+    await integration_session.commit()
+
+    with patch("routstr.payment.models.sats_usd_price", return_value=1e-6):
+        await reinitialize_upstreams()
+
+    with patch("routstr.payment.cost_calculation.sats_usd_price", return_value=1e-6):
+        cost = await calculate_cost(
+            {
+                "model": "cache-priced-model",
+                "usage": {
+                    "prompt_tokens": 1000,
+                    "completion_tokens": 100,
+                    "prompt_tokens_details": {"cached_tokens": 800},
+                },
+            },
+            max_cost=1_000_000,
+        )
+
+    assert isinstance(cost, CostData)
+    # prompt 1.4e-7 USD/token -> 0.14 sats/token -> 140 msats/token
+    # cache 1.25e-9 USD/token -> 0.00125 sats/token -> 1.25 msats/token
+    # completion 2.8e-7 USD/token -> 0.28 sats/token -> 280 msats/token
+    assert cost.input_tokens == 200
+    assert cost.cache_read_input_tokens == 800
+    assert cost.cache_read_msats == 1000
+    assert cost.input_msats == 29000
+    assert cost.output_msats == 28000
+    assert cost.total_msats == 57000

tests/integration/test_admin_nsec_endpoint.py

@@ -0,0 +1,113 @@
+"""Tests for the admin nsec rotation endpoint (issue #553).
+
+The Nostr identity is a secret: it lives encrypted in the Secret store, never in
+the settings blob, so it cannot be set through the general settings PATCH. This
+dedicated endpoint is the supported way to set/rotate/clear it — it encrypts the
+key at rest, updates the live runtime identity (so signing picks it up without a
+restart), and derives the npub. Invalid keys are rejected.
+"""
+
+from __future__ import annotations
+
+import secrets
+import time
+from collections.abc import AsyncGenerator
+
+import pytest
+import pytest_asyncio
+from httpx import AsyncClient
+
+from routstr.core import vault
+from routstr.core.admin import admin_sessions
+from routstr.core.db import AsyncSession, get_secret
+from routstr.core.settings import derive_npub_from_nsec, settings
+
+# A valid 64-char hex private key (accepted by nsec_to_keypair, as in bootstrap).
+NSEC_HEX = "1" * 64
+
+
+@pytest_asyncio.fixture
+async def admin_client(
+    integration_client: AsyncClient,
+) -> AsyncGenerator[AsyncClient, None]:
+    """An integration_client pre-authenticated with an admin session token."""
+    token = secrets.token_urlsafe(24)
+    admin_sessions[token] = int(time.time()) + 3600
+    integration_client.headers["Authorization"] = f"Bearer {token}"
+    yield integration_client
+    admin_sessions.pop(token, None)
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_update_nsec_stores_encrypted_and_derives_npub(
+    admin_client: AsyncClient,
+    integration_session: AsyncSession,
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    monkeypatch.setattr(settings, "nsec", "")
+    monkeypatch.setattr(settings, "npub", "")
+
+    resp = await admin_client.patch("/admin/api/nsec", json={"nsec": NSEC_HEX})
+    assert resp.status_code == 200
+
+    expected_npub = derive_npub_from_nsec(NSEC_HEX)
+    assert resp.json() == {"ok": True, "npub": expected_npub}
+
+    # Stored encrypted at rest, decryptable back to the original key.
+    integration_session.expunge_all()
+    secret = await get_secret(integration_session)
+    assert secret.encrypted_nsec is not None
+    assert vault.is_encrypted(secret.encrypted_nsec)
+    assert vault.decrypt(secret.encrypted_nsec) == NSEC_HEX
+
+    # Live runtime identity updated so Nostr signing reflects it without restart.
+    assert settings.nsec == NSEC_HEX
+    assert settings.npub == expected_npub
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_update_nsec_rejects_invalid_key(
+    admin_client: AsyncClient,
+    integration_session: AsyncSession,
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    monkeypatch.setattr(settings, "nsec", "")
+    monkeypatch.setattr(settings, "npub", "")
+
+    resp = await admin_client.patch(
+        "/admin/api/nsec", json={"nsec": "not-a-real-nsec"}
+    )
+    assert resp.status_code == 400
+
+    # Nothing stored, live identity untouched.
+    integration_session.expunge_all()
+    secret = await get_secret(integration_session)
+    assert secret.encrypted_nsec is None
+    assert settings.nsec == ""
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_update_nsec_clears_identity_with_empty_value(
+    admin_client: AsyncClient,
+    integration_session: AsyncSession,
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    # Start from a node that has an identity...
+    monkeypatch.setattr(settings, "nsec", "")
+    monkeypatch.setattr(settings, "npub", "")
+    set_resp = await admin_client.patch("/admin/api/nsec", json={"nsec": NSEC_HEX})
+    assert set_resp.status_code == 200
+
+    # ...then clear it.
+    clear_resp = await admin_client.patch("/admin/api/nsec", json={"nsec": ""})
+    assert clear_resp.status_code == 200
+    assert clear_resp.json() == {"ok": True, "npub": ""}
+
+    integration_session.expunge_all()
+    secret = await get_secret(integration_session)
+    assert secret.encrypted_nsec is None
+    assert settings.nsec == ""
+    assert settings.npub == ""

tests/integration/test_admin_settings_endpoint.py

@@ -0,0 +1,82 @@
+"""Tests for the admin settings endpoint's handling of secrets (issue #553).
+
+``admin_password`` is no longer a settings field (it lives only as a one-way
+hash in the Secret store), so it must never appear in the GET/PATCH payloads.
+``nsec`` (in-memory at runtime) and ``upstream_api_key`` (still in the settings
+blob) are both redacted on read and ignored on write — they cannot be set
+through the general settings endpoint, only through their dedicated paths.
+"""
+
+from __future__ import annotations
+
+import secrets
+import time
+from collections.abc import AsyncGenerator
+
+import pytest
+import pytest_asyncio
+from httpx import AsyncClient
+
+from routstr.core.admin import admin_sessions
+from routstr.core.db import AsyncSession
+from routstr.core.settings import SettingsService, settings
+
+
+@pytest_asyncio.fixture
+async def admin_client(
+    integration_client: AsyncClient,
+) -> AsyncGenerator[AsyncClient, None]:
+    """An integration_client pre-authenticated with an admin session token."""
+    token = secrets.token_urlsafe(24)
+    admin_sessions[token] = int(time.time()) + 3600
+    integration_client.headers["Authorization"] = f"Bearer {token}"
+    yield integration_client
+    admin_sessions.pop(token, None)
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_get_settings_omits_admin_password_and_redacts_secrets(
+    admin_client: AsyncClient, monkeypatch: pytest.MonkeyPatch
+) -> None:
+    monkeypatch.setattr(settings, "nsec", "nsec-secret")
+    monkeypatch.setattr(settings, "upstream_api_key", "sk-secret")
+
+    resp = await admin_client.get("/admin/api/settings")
+    assert resp.status_code == 200
+
+    data = resp.json()
+    assert "admin_password" not in data
+    assert data["nsec"] == "[REDACTED]"
+    assert data["upstream_api_key"] == "[REDACTED]"
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_patch_settings_ignores_secret_fields(
+    admin_client: AsyncClient,
+    integration_session: AsyncSession,
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    # The PATCH path persists through SettingsService, which needs an
+    # initialized current snapshot and a settings row in the shared test DB.
+    await SettingsService.initialize(integration_session)
+    monkeypatch.setattr(settings, "nsec", "original-nsec")
+
+    resp = await admin_client.patch(
+        "/admin/api/settings",
+        json={
+            "name": "Renamed",
+            "nsec": "attacker-nsec",
+            "upstream_api_key": "attacker-key",
+            "admin_password": "attacker-pw",
+        },
+    )
+    assert resp.status_code == 200
+
+    data = resp.json()
+    assert data["name"] == "Renamed"
+    assert "admin_password" not in data
+    assert data["nsec"] == "[REDACTED]"
+    # The live secret was not overwritten through the general settings endpoint.
+    assert settings.nsec == "original-nsec"

tests/integration/test_balance_negative_on_cost_overrun.py

@@ -15,6 +15,7 @@ from unittest.mock import patch
 import pytest
 from sqlmodel.ext.asyncio.session import AsyncSession
 
+from routstr.auth import ReservationSnapshot
 from routstr.core.db import ApiKey
 from routstr.payment.cost_calculation import CostData
 
@@ -23,7 +24,7 @@ def _make_key(balance: int, reserved: int) -> ApiKey:
     return ApiKey(
         hashed_key=f"test_{uuid.uuid4().hex}",
         balance=balance,
-        reserved_balance=reserved,
+        reserved_balance=0,
         total_spent=0,
         total_requests=1,
     )
@@ -75,6 +76,8 @@ async def test_balance_never_negative_when_cost_exceeds_reservation(
     key = _make_key(balance=deducted_max_cost, reserved=deducted_max_cost)
     integration_session.add(key)
     await integration_session.commit()
+    from routstr.auth import pay_for_request
+    await pay_for_request(key, deducted_max_cost, integration_session)
 
     response_data = {"model": "test-model", "usage": {"prompt_tokens": 100, "completion_tokens": 100}}
 
@@ -82,7 +85,9 @@ async def test_balance_never_negative_when_cost_exceeds_reservation(
         "routstr.auth.calculate_cost",
         return_value=_cost_data(actual_token_cost),
     ):
-        await adjust_payment_for_tokens(key, response_data, integration_session, deducted_max_cost)
+        await adjust_payment_for_tokens(
+            key, response_data, integration_session, deducted_max_cost, None, None
+        )
 
     await _refresh(integration_session, key)
 
@@ -109,6 +114,8 @@ async def test_balance_floor_at_zero_on_overrun(
     key = _make_key(balance=500, reserved=500)
     integration_session.add(key)
     await integration_session.commit()
+    from routstr.auth import pay_for_request
+    await pay_for_request(key, deducted_max_cost, integration_session)
 
     response_data = {"model": "test-model", "usage": {"prompt_tokens": 50, "completion_tokens": 50}}
 
@@ -116,7 +123,9 @@ async def test_balance_floor_at_zero_on_overrun(
         "routstr.auth.calculate_cost",
         return_value=_cost_data(actual_token_cost),
     ):
-        await adjust_payment_for_tokens(key, response_data, integration_session, deducted_max_cost)
+        await adjust_payment_for_tokens(
+            key, response_data, integration_session, deducted_max_cost, None, None
+        )
 
     await _refresh(integration_session, key)
 
@@ -148,6 +157,8 @@ async def test_full_cost_charged_when_balance_sufficient_for_overrun(
     key = _make_key(balance=2000, reserved=990)
     integration_session.add(key)
     await integration_session.commit()
+    from routstr.auth import pay_for_request
+    await pay_for_request(key, deducted_max_cost, integration_session)
 
     response_data = {"model": "test-model", "usage": {"prompt_tokens": 100, "completion_tokens": 100}}
 
@@ -155,7 +166,9 @@ async def test_full_cost_charged_when_balance_sufficient_for_overrun(
         "routstr.auth.calculate_cost",
         return_value=_cost_data(actual_token_cost),
     ):
-        await adjust_payment_for_tokens(key, response_data, integration_session, deducted_max_cost)
+        await adjust_payment_for_tokens(
+            key, response_data, integration_session, deducted_max_cost, None, None
+        )
 
     await _refresh(integration_session, key)
 
@@ -184,7 +197,11 @@ async def test_concurrent_cost_overruns_never_negative(
     """Concurrent finalization with cost overruns must never produce negative balance."""
     import asyncio
 
-    from routstr.auth import adjust_payment_for_tokens, pay_for_request
+    from routstr.auth import (
+        adjust_payment_for_tokens,
+        get_reservation_snapshot,
+        pay_for_request,
+    )
     from routstr.core.db import create_session
 
     deducted_max_cost = 990
@@ -210,12 +227,14 @@ async def test_concurrent_cost_overruns_never_negative(
     async with create_session() as session:
         key_to_reserve = await session.get(ApiKey, key_hash)
         assert key_to_reserve is not None
+        reservations = []
         for _ in range(n_requests):
             await pay_for_request(key_to_reserve, deducted_max_cost, session)
+            reservations.append(await get_reservation_snapshot(key_to_reserve, session))
             await session.refresh(key_to_reserve)
 
     # Now finalize all concurrently with cost overrun
-    async def finalize() -> None:
+    async def finalize(reservation: ReservationSnapshot) -> None:
         response_data = {
             "model": "test-model",
             "usage": {"prompt_tokens": 100, "completion_tokens": 100},
@@ -223,15 +242,22 @@ async def test_concurrent_cost_overruns_never_negative(
         async with create_session() as session:
             fresh_key = await session.get(ApiKey, key_hash)
             assert fresh_key is not None
-            with patch(
-                "routstr.auth.calculate_cost",
-                return_value=_cost_data(actual_token_cost),
-            ):
-                await adjust_payment_for_tokens(
-                    fresh_key, response_data, session, deducted_max_cost
-                )
+            await adjust_payment_for_tokens(
+                fresh_key,
+                response_data,
+                session,
+                deducted_max_cost,
+                reservation_snapshot=reservation,
+            )
 
-    await asyncio.gather(*[finalize() for _ in range(n_requests)])
+    # Patch once around the gather: entering the same patch target from
+    # concurrent tasks un-patches in the wrong order and leaks the mock into
+    # every later test in the session.
+    with patch(
+        "routstr.auth.calculate_cost",
+        return_value=_cost_data(actual_token_cost),
+    ):
+        await asyncio.gather(*(finalize(r) for r in reservations))
 
     async with create_session() as session:
         final_key = await session.get(ApiKey, key_hash)
@@ -272,6 +298,8 @@ async def test_zero_free_balance_overrun_is_safe(
     key = _make_key(balance=1000, reserved=1000)
     integration_session.add(key)
     await integration_session.commit()
+    from routstr.auth import pay_for_request
+    await pay_for_request(key, deducted_max_cost, integration_session)
 
     response_data = {"model": "test-model", "usage": {"prompt_tokens": 50, "completion_tokens": 100}}
 
@@ -279,7 +307,9 @@ async def test_zero_free_balance_overrun_is_safe(
         "routstr.auth.calculate_cost",
         return_value=_cost_data(actual_token_cost),
     ):
-        await adjust_payment_for_tokens(key, response_data, integration_session, deducted_max_cost)
+        await adjust_payment_for_tokens(
+            key, response_data, integration_session, deducted_max_cost, None, None
+        )
 
     await _refresh(integration_session, key)
 
@@ -308,7 +338,11 @@ async def test_parallel_requests_no_free_inference(
     """Second parallel finalization must be charged even when first depleted free balance."""
     import asyncio
 
-    from routstr.auth import adjust_payment_for_tokens
+    from routstr.auth import (
+        adjust_payment_for_tokens,
+        get_reservation_snapshot,
+        pay_for_request,
+    )
     from routstr.core.db import create_session
 
     deducted_max_cost = 100
@@ -329,14 +363,18 @@ async def test_parallel_requests_no_free_inference(
         key = ApiKey(
             hashed_key=key_hash,
             balance=starting_balance,
-            reserved_balance=deducted_max_cost * 2,  # both slots pre-reserved
+            reserved_balance=0,
             total_spent=0,
             total_requests=2,
         )
         session.add(key)
         await session.commit()
+        reservations = []
+        for _ in range(2):
+            await pay_for_request(key, deducted_max_cost, session)
+            reservations.append(await get_reservation_snapshot(key, session))
 
-    async def finalize() -> None:
+    async def finalize(reservation: ReservationSnapshot) -> None:
         response_data = {
             "model": "test-model",
             "usage": {"prompt_tokens": 50, "completion_tokens": 100},
@@ -344,15 +382,22 @@ async def test_parallel_requests_no_free_inference(
         async with create_session() as session:
             fresh_key = await session.get(ApiKey, key_hash)
             assert fresh_key is not None
-            with patch(
-                "routstr.auth.calculate_cost",
-                return_value=_cost_data(actual_token_cost),
-            ):
-                await adjust_payment_for_tokens(
-                    fresh_key, response_data, session, deducted_max_cost
-                )
+            await adjust_payment_for_tokens(
+                fresh_key,
+                response_data,
+                session,
+                deducted_max_cost,
+                reservation_snapshot=reservation,
+            )
 
-    await asyncio.gather(finalize(), finalize())
+    # Patch once around the gather: entering the same patch target from two
+    # concurrent tasks un-patches in the wrong order and leaks the mock into
+    # every later test in the session.
+    with patch(
+        "routstr.auth.calculate_cost",
+        return_value=_cost_data(actual_token_cost),
+    ):
+        await asyncio.gather(*(finalize(r) for r in reservations))
 
     async with create_session() as session:
         final_key = await session.get(ApiKey, key_hash)

tests/integration/test_child_keys.py

@@ -77,7 +77,7 @@ async def test_child_key_flow(integration_session: AsyncSession) -> None:
 
     try:
         adjustment = await adjust_payment_for_tokens(
-            child_key_db, response_data, integration_session, 500
+            child_key_db, response_data, integration_session, 500, None, None
         )
         assert adjustment["total_msats"] == 400
 

tests/integration/test_failover_billing.py

@@ -0,0 +1,666 @@
+"""Failover requests are billed and forwarded as the provider that served them.
+
+Covers the whole-system settlement path when two enabled providers expose the
+same model under different spellings and prices: the routing winner fails with
+a 502, the fallback provider serves, and the response must be billed at the
+fallback's configured rate, carry the fallback's model id in the forwarded
+request body, and echo the fallback's model id to the client.
+"""
+
+import json
+from typing import Any, AsyncGenerator
+from unittest.mock import patch
+
+import httpx
+import pytest
+from httpx import AsyncClient
+from sqlmodel import select
+from sqlmodel.ext.asyncio.session import AsyncSession
+
+from routstr.core.db import ApiKey, ReservationRelease
+from routstr.payment.models import Architecture, Model, Pricing
+from routstr.proxy import refresh_model_maps
+from routstr.upstream.base import BaseUpstreamProvider
+
+CHEAP_BASE_URL = "https://cheap.example.com/v1"
+EXPENSIVE_BASE_URL = "https://expensive.example.com/v1"
+THIRD_BASE_URL = "https://third.example.com/v1"
+
+
+def _make_model(
+    model_id: str,
+    prompt_sats: float,
+    completion_sats: float,
+    max_cost: float = 50.0,
+) -> Model:
+    """Build a model whose USD and sats pricing rank consistently."""
+    return Model(
+        id=model_id,
+        name=model_id,
+        created=1,
+        description="test model",
+        context_length=8192,
+        architecture=Architecture(
+            modality="text",
+            input_modalities=["text"],
+            output_modalities=["text"],
+            tokenizer="gpt",
+            instruct_type=None,
+        ),
+        pricing=Pricing(
+            prompt=prompt_sats, completion=completion_sats, max_cost=max_cost
+        ),
+        sats_pricing=Pricing(
+            prompt=prompt_sats, completion=completion_sats, max_cost=max_cost
+        ),
+    )
+
+
+class _StaticProvider(BaseUpstreamProvider):
+    """Upstream provider with a fixed model catalog and no remote refresh."""
+
+    def __init__(self, base_url: str, api_key: str, fee: float, model: Model) -> None:
+        super().__init__(base_url, api_key, fee)
+        self.provider_type = "custom"
+        self._static_model = model
+
+    def get_cached_models(self) -> list[Model]:
+        return [self._static_model]
+
+    async def refresh_models_cache(self) -> None:
+        pass
+
+
+async def _install_providers(
+    providers: list[_StaticProvider],
+) -> AsyncGenerator[None, None]:
+    """Install providers into the routing maps, restoring the originals after."""
+    from routstr import proxy
+
+    original_upstreams = proxy.get_upstreams()
+    with patch("routstr.proxy._upstreams", providers):
+        await refresh_model_maps()
+        yield
+    with patch("routstr.proxy._upstreams", original_upstreams):
+        await refresh_model_maps()
+
+
+@pytest.fixture
+async def dual_provider_maps(
+    patched_db_engine: None,
+) -> AsyncGenerator[tuple[_StaticProvider, _StaticProvider], None]:
+    """Two same-tail providers under different spellings and prices."""
+    cheap = _StaticProvider(
+        CHEAP_BASE_URL,
+        "key-cheap",
+        1.0,
+        _make_model("prova/dual-model", 0.001, 0.002),
+    )
+    expensive = _StaticProvider(
+        EXPENSIVE_BASE_URL,
+        "key-expensive",
+        1.0,
+        _make_model("provb/dual-model", 0.005, 0.010, max_cost=100.0),
+    )
+    async for _ in _install_providers([cheap, expensive]):
+        yield cheap, expensive
+
+
+def _upstream_response(request: httpx.Request) -> httpx.Response:
+    """502 from the cheap (winning) provider; a served completion elsewhere."""
+    if request.url.host == "cheap.example.com":
+        return httpx.Response(
+            502,
+            content=json.dumps({"error": {"message": "bad gateway"}}).encode(),
+            headers={"content-type": "application/json"},
+        )
+    body = {
+        "id": "chatcmpl-served",
+        "object": "chat.completion",
+        "created": 1,
+        "model": "dual-model",
+        "choices": [
+            {
+                "index": 0,
+                "message": {"role": "assistant", "content": "hi"},
+                "finish_reason": "stop",
+            }
+        ],
+        "usage": {
+            "prompt_tokens": 1000,
+            "completion_tokens": 500,
+            "total_tokens": 1500,
+        },
+    }
+    return httpx.Response(
+        200,
+        content=json.dumps(body).encode(),
+        headers={"content-type": "application/json"},
+    )
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_failover_serve_billed_at_serving_providers_rate(
+    authenticated_client: AsyncClient,
+    dual_provider_maps: tuple[_StaticProvider, _StaticProvider],
+    integration_session: AsyncSession,
+) -> None:
+    """A fallback serve is billed at the fallback's price, not the winner's.
+
+    The cheap provider ranks first for the shared tail; it 502s and the
+    expensive provider serves 1000 input + 500 output tokens. At the serving
+    provider's sats pricing (0.005/0.010 sats per token) that is 10_000 msats;
+    at the winner's (0.001/0.002) it would be 2_000 msats.
+    """
+    sent_requests: list[httpx.Request] = []
+
+    # Patch the network transport (not AsyncClient.send) so the in-process
+    # ASGI test client is untouched and only the proxy's upstream hop is mocked.
+    async def fake_transport(
+        request: httpx.Request, *args: Any, **kwargs: Any
+    ) -> httpx.Response:
+        sent_requests.append(request)
+        return _upstream_response(request)
+
+    with (
+        patch(
+            "httpx.AsyncHTTPTransport.handle_async_request",
+            side_effect=fake_transport,
+        ),
+        # cost_calculation binds sats_usd_price at import time, so the price
+        # patch in the app fixture does not reach it; patch its own binding.
+        patch(
+            "routstr.payment.cost_calculation.sats_usd_price",
+            return_value=0.0005,
+        ),
+    ):
+        response = await authenticated_client.post(
+            "/v1/chat/completions",
+            json={
+                "model": "dual-model",
+                "messages": [{"role": "user", "content": "hello"}],
+            },
+        )
+
+    assert response.status_code == 200
+    payload = response.json()
+
+    # Both providers were attempted, cheapest first.
+    assert [r.url.host for r in sent_requests] == [
+        "cheap.example.com",
+        "expensive.example.com",
+    ]
+
+    # The fallback must be asked for ITS OWN model spelling, not the winner's.
+    forwarded_body = json.loads(sent_requests[1].content)
+    assert forwarded_body["model"] == "provb/dual-model"
+
+    # The response echo names the model that actually served.
+    assert payload["model"] == "provb/dual-model"
+
+    # Billed at the serving provider's rate: 1000/1000*5000 + 500/1000*10000.
+    assert payload["cost"]["total_msats"] == 10_000
+
+    # The fallback's larger max-cost envelope requires a replacement
+    # reservation. The failed candidate is released, the serving candidate is
+    # charged, and no request-owned reservation remains active.
+    key_hash = authenticated_client._test_api_key.removeprefix("sk-")  # type: ignore[attr-defined]
+    records = (
+        await integration_session.exec(
+            select(ReservationRelease).where(ReservationRelease.key_hash == key_hash)
+        )
+    ).all()
+    assert sorted(record.status for record in records) == ["charged", "released"]
+
+
+@pytest.fixture
+async def same_id_provider_maps(
+    patched_db_engine: None,
+) -> AsyncGenerator[None, None]:
+    """Two providers exposing the IDENTICAL model id at different prices."""
+    cheap = _StaticProvider(
+        CHEAP_BASE_URL,
+        "key-cheap",
+        1.0,
+        _make_model("dual-model", 0.001, 0.002),
+    )
+    expensive = _StaticProvider(
+        EXPENSIVE_BASE_URL,
+        "key-expensive",
+        1.0,
+        _make_model("dual-model", 0.005, 0.010),
+    )
+    async for _ in _install_providers([cheap, expensive]):
+        yield
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_same_id_failover_settles_at_serving_price(
+    authenticated_client: AsyncClient,
+    same_id_provider_maps: None,
+) -> None:
+    """Settlement must not re-derive pricing from the response's model string.
+
+    Both providers expose the exact same model id, so the forwarded body is
+    identical either way — the only observable difference is the settled
+    amount. The response's model string resolves to the alias winner (cheap),
+    but the expensive provider served, so the bill must be 10_000 msats, not
+    the winner's 2_000.
+    """
+    sent_requests: list[httpx.Request] = []
+
+    async def fake_transport(
+        request: httpx.Request, *args: Any, **kwargs: Any
+    ) -> httpx.Response:
+        sent_requests.append(request)
+        return _upstream_response(request)
+
+    with (
+        patch(
+            "httpx.AsyncHTTPTransport.handle_async_request",
+            side_effect=fake_transport,
+        ),
+        patch(
+            "routstr.payment.cost_calculation.sats_usd_price",
+            return_value=0.0005,
+        ),
+    ):
+        response = await authenticated_client.post(
+            "/v1/chat/completions",
+            json={
+                "model": "dual-model",
+                "messages": [{"role": "user", "content": "hello"}],
+            },
+        )
+
+    assert response.status_code == 200
+    assert [r.url.host for r in sent_requests] == [
+        "cheap.example.com",
+        "expensive.example.com",
+    ]
+    assert response.json()["cost"]["total_msats"] == 10_000
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_version_suffixed_model_id_routes(
+    authenticated_client: AsyncClient,
+    same_id_provider_maps: None,
+) -> None:
+    """A version-suffixed request (``…-YYYYMMDD``) routes to the base model.
+
+    Model resolution stripped the suffix but the provider lookup did not, so
+    such requests resolved a model yet found no provider and 400'd. With the
+    unified candidate lookup the strip applies to both.
+    """
+
+    async def fake_transport(
+        request: httpx.Request, *args: Any, **kwargs: Any
+    ) -> httpx.Response:
+        return _upstream_response(request)
+
+    with (
+        patch(
+            "httpx.AsyncHTTPTransport.handle_async_request",
+            side_effect=fake_transport,
+        ),
+        patch(
+            "routstr.payment.cost_calculation.sats_usd_price",
+            return_value=0.0005,
+        ),
+    ):
+        response = await authenticated_client.post(
+            "/v1/chat/completions",
+            json={
+                "model": "dual-model-20260101",
+                "messages": [{"role": "user", "content": "hello"}],
+            },
+        )
+
+    assert response.status_code == 200
+
+
+@pytest.fixture
+async def fee_split_provider_maps(
+    patched_db_engine: None,
+) -> AsyncGenerator[None, None]:
+    """Same-tail providers whose fees differ; the serving one charges 1.5x."""
+    cheap = _StaticProvider(
+        CHEAP_BASE_URL,
+        "key-cheap",
+        1.0,
+        _make_model("dual-model", 0.001, 0.002),
+    )
+    expensive = _StaticProvider(
+        EXPENSIVE_BASE_URL,
+        "key-expensive",
+        1.5,
+        _make_model("dual-model", 0.005, 0.010),
+    )
+    async for _ in _install_providers([cheap, expensive]):
+        yield
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_usd_cost_serve_carries_serving_providers_fee(
+    authenticated_client: AsyncClient,
+    fee_split_provider_maps: None,
+) -> None:
+    """The USD-cost billing path applies the SERVING provider's fee.
+
+    The upstream that serves reports ``usage.cost`` in USD, so billing goes
+    through the USD-cost path where the provider fee is applied explicitly.
+    The serving provider's fee is 1.5; the alias winner's is 1.0. At 0.001 USD
+    reported cost and 0.0005 USD/sat: 0.001 * 1.5 / 0.0005 = 3 sats = 3000
+    msats (fee 1.0 would give 2000).
+    """
+    sent_requests: list[httpx.Request] = []
+
+    def usd_cost_response(request: httpx.Request) -> httpx.Response:
+        if request.url.host == "cheap.example.com":
+            return httpx.Response(
+                502,
+                content=json.dumps({"error": {"message": "bad gateway"}}).encode(),
+                headers={"content-type": "application/json"},
+            )
+        body = {
+            "id": "chatcmpl-usd",
+            "object": "chat.completion",
+            "created": 1,
+            "model": "dual-model",
+            "choices": [
+                {
+                    "index": 0,
+                    "message": {"role": "assistant", "content": "hi"},
+                    "finish_reason": "stop",
+                }
+            ],
+            "usage": {
+                "prompt_tokens": 100,
+                "completion_tokens": 50,
+                "total_tokens": 150,
+                "cost": 0.001,
+            },
+        }
+        return httpx.Response(
+            200,
+            content=json.dumps(body).encode(),
+            headers={"content-type": "application/json"},
+        )
+
+    async def fake_transport(
+        request: httpx.Request, *args: Any, **kwargs: Any
+    ) -> httpx.Response:
+        sent_requests.append(request)
+        return usd_cost_response(request)
+
+    with (
+        patch(
+            "httpx.AsyncHTTPTransport.handle_async_request",
+            side_effect=fake_transport,
+        ),
+        patch(
+            "routstr.payment.cost_calculation.sats_usd_price",
+            return_value=0.0005,
+        ),
+    ):
+        response = await authenticated_client.post(
+            "/v1/chat/completions",
+            json={
+                "model": "dual-model",
+                "messages": [{"role": "user", "content": "hello"}],
+            },
+        )
+
+    assert response.status_code == 200
+    assert [r.url.host for r in sent_requests] == [
+        "cheap.example.com",
+        "expensive.example.com",
+    ]
+    assert response.json()["cost"]["total_msats"] == 3_000
+
+
+@pytest.fixture
+async def envelope_split_provider_maps(
+    patched_db_engine: None,
+) -> AsyncGenerator[None, None]:
+    """Same-id providers where the fallback's max cost dwarfs the key balance."""
+    cheap = _StaticProvider(
+        CHEAP_BASE_URL,
+        "key-cheap",
+        1.0,
+        _make_model("dual-model", 0.001, 0.002, max_cost=50.0),
+    )
+    expensive = _StaticProvider(
+        EXPENSIVE_BASE_URL,
+        "key-expensive",
+        1.0,
+        _make_model("dual-model", 0.005, 0.010, max_cost=20_000.0),
+    )
+    async for _ in _install_providers([cheap, expensive]):
+        yield
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_failover_beyond_balance_envelope_is_rejected(
+    authenticated_client: AsyncClient,
+    envelope_split_provider_maps: None,
+) -> None:
+    """A fallback whose max-cost envelope exceeds the balance is not served.
+
+    Admission and reservation are sized to the best-ranked candidate's max
+    cost. When that candidate fails and the next one's envelope exceeds the
+    key's balance, serving it could settle far beyond what admission allowed,
+    so the request must be rejected (as it would be if the pricier candidate
+    were ranked first) instead of forwarded.
+    """
+    sent_requests: list[httpx.Request] = []
+
+    async def fake_transport(
+        request: httpx.Request, *args: Any, **kwargs: Any
+    ) -> httpx.Response:
+        sent_requests.append(request)
+        return _upstream_response(request)
+
+    with (
+        patch(
+            "httpx.AsyncHTTPTransport.handle_async_request",
+            side_effect=fake_transport,
+        ),
+        patch(
+            "routstr.payment.cost_calculation.sats_usd_price",
+            return_value=0.0005,
+        ),
+    ):
+        response = await authenticated_client.post(
+            "/v1/chat/completions",
+            json={
+                "model": "dual-model",
+                "messages": [{"role": "user", "content": "hello"}],
+            },
+        )
+
+    # The 20_000-sat envelope exceeds the key's 10_000-sat balance: the
+    # fallback must be rejected before its upstream is ever contacted.
+    assert response.status_code == 402
+    assert [r.url.host for r in sent_requests] == ["cheap.example.com"]
+
+
+@pytest.fixture
+async def three_candidate_child_maps(
+    patched_db_engine: None,
+) -> AsyncGenerator[None, None]:
+    """Second candidate cannot fit the child limit; third restores and serves."""
+    first = _StaticProvider(
+        CHEAP_BASE_URL,
+        "key-first",
+        1.0,
+        _make_model("dual-model", 0.001, 0.002, max_cost=50.0),
+    )
+    too_large = _StaticProvider(
+        EXPENSIVE_BASE_URL,
+        "key-too-large",
+        1.0,
+        _make_model("dual-model", 0.002, 0.003, max_cost=100.0),
+    )
+    third = _StaticProvider(
+        THIRD_BASE_URL,
+        "key-third",
+        1.0,
+        _make_model("dual-model", 0.003, 0.004, max_cost=50.0),
+    )
+    async for _ in _install_providers([first, too_large, third]):
+        yield
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_child_failover_rolls_back_failed_larger_reserve_before_restoring(
+    authenticated_client: AsyncClient,
+    three_candidate_child_maps: None,
+    integration_session: AsyncSession,
+) -> None:
+    """A failed child guard cannot leak its parent update into restoration."""
+    key_hash = authenticated_client._test_api_key.removeprefix("sk-")  # type: ignore[attr-defined]
+    child = await integration_session.get(ApiKey, key_hash)
+    assert child is not None
+    parent = ApiKey(hashed_key="failover-parent", balance=10_000_000)
+    child.parent_key_hash = parent.hashed_key
+    child.balance_limit = 75_000
+    integration_session.add(parent)
+    integration_session.add(child)
+    await integration_session.commit()
+
+    sent_requests: list[httpx.Request] = []
+
+    async def fake_transport(
+        request: httpx.Request, *args: Any, **kwargs: Any
+    ) -> httpx.Response:
+        sent_requests.append(request)
+        return _upstream_response(request)
+
+    with (
+        patch(
+            "httpx.AsyncHTTPTransport.handle_async_request",
+            side_effect=fake_transport,
+        ),
+        patch(
+            "routstr.payment.cost_calculation.sats_usd_price",
+            return_value=0.0005,
+        ),
+    ):
+        response = await authenticated_client.post(
+            "/v1/chat/completions",
+            json={
+                "model": "dual-model",
+                "messages": [{"role": "user", "content": "hello"}],
+            },
+        )
+
+    assert response.status_code == 200
+    # The 100-sat candidate is rejected before forwarding; the third serves.
+    assert [request.url.host for request in sent_requests] == [
+        "cheap.example.com",
+        "third.example.com",
+    ]
+
+    await integration_session.refresh(parent)
+    await integration_session.refresh(child)
+    assert parent.reserved_balance == 0
+    assert child.reserved_balance == 0
+    assert parent.total_spent == response.json()["cost"]["total_msats"]
+
+    records = (
+        await integration_session.exec(
+            select(ReservationRelease).where(ReservationRelease.key_hash == key_hash)
+        )
+    ).all()
+    assert len(records) == 2
+    assert sorted(record.status for record in records) == ["charged", "released"]
+    assert len({record.reserved_msats for record in records}) == 1
+    assert all(record.status != "active" for record in records)
+
+
+@pytest.fixture
+async def raised_envelope_provider_maps(
+    patched_db_engine: None,
+) -> AsyncGenerator[None, None]:
+    """Same-id providers where the fallback needs a larger, affordable reserve."""
+    cheap = _StaticProvider(
+        CHEAP_BASE_URL,
+        "key-cheap",
+        1.0,
+        _make_model("dual-model", 0.001, 0.002, max_cost=50.0),
+    )
+    expensive = _StaticProvider(
+        EXPENSIVE_BASE_URL,
+        "key-expensive",
+        1.0,
+        _make_model("dual-model", 0.005, 0.010, max_cost=100.0),
+    )
+    async for _ in _install_providers([cheap, expensive]):
+        yield
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_failover_reserves_serving_candidates_envelope(
+    authenticated_client: AsyncClient,
+    raised_envelope_provider_maps: None,
+    integration_session: AsyncSession,
+) -> None:
+    """An affordable pricier fallback is re-reserved, served, and billed.
+
+    The fallback's max cost (100 sats) exceeds the winner's (50 sats) but fits
+    the key's balance, so the reservation is raised to the serving candidate's
+    envelope and the request completes, billed at the serving rate with the
+    unused reserve refunded.
+    """
+    sent_requests: list[httpx.Request] = []
+
+    async def fake_transport(
+        request: httpx.Request, *args: Any, **kwargs: Any
+    ) -> httpx.Response:
+        sent_requests.append(request)
+        return _upstream_response(request)
+
+    with (
+        patch(
+            "httpx.AsyncHTTPTransport.handle_async_request",
+            side_effect=fake_transport,
+        ),
+        patch(
+            "routstr.payment.cost_calculation.sats_usd_price",
+            return_value=0.0005,
+        ),
+    ):
+        response = await authenticated_client.post(
+            "/v1/chat/completions",
+            json={
+                "model": "dual-model",
+                "messages": [{"role": "user", "content": "hello"}],
+            },
+        )
+
+    assert response.status_code == 200
+    assert [r.url.host for r in sent_requests] == [
+        "cheap.example.com",
+        "expensive.example.com",
+    ]
+    assert response.json()["cost"]["total_msats"] == 10_000
+
+    key_hash = authenticated_client._test_api_key.removeprefix("sk-")  # type: ignore[attr-defined]
+    records = (
+        await integration_session.exec(
+            select(ReservationRelease).where(ReservationRelease.key_hash == key_hash)
+        )
+    ).all()
+    assert len(records) == 2
+    released = next(record for record in records if record.status == "released")
+    charged = next(record for record in records if record.status == "charged")
+    assert charged.reserved_msats > released.reserved_msats
+    assert all(record.status != "active" for record in records)

tests/integration/test_free_response_stale_reservation.py

@@ -0,0 +1,165 @@
+"""Regression tests for charging after stale reservation cleanup."""
+
+import time
+import uuid
+from unittest.mock import patch
+
+import pytest
+from sqlmodel.ext.asyncio.session import AsyncSession
+
+from routstr.core.db import ApiKey
+from routstr.payment.cost_calculation import CostData
+
+
+def _make_key(balance: int, reserved: int) -> ApiKey:
+    return ApiKey(
+        hashed_key=f"test_{uuid.uuid4().hex}",
+        balance=balance,
+        reserved_balance=reserved,
+        total_spent=0,
+        total_requests=1,
+    )
+
+
+def _cost_data(total_msats: int) -> CostData:
+    return CostData(
+        base_msats=0,
+        input_msats=total_msats // 2,
+        output_msats=total_msats - total_msats // 2,
+        total_msats=total_msats,
+        total_usd=0.0,
+        input_tokens=100,
+        output_tokens=100,
+    )
+
+
+@pytest.mark.asyncio
+async def test_overrun_charges_after_reservation_swept(
+    integration_session: AsyncSession,
+) -> None:
+    """Overrun finalize must charge even when the reservation was already released."""
+    from routstr.auth import adjust_payment_for_tokens, pay_for_request
+
+    deducted_max_cost = 990  # discounted reservation
+    actual_token_cost = 1000  # actual cost overruns the reservation
+
+    # Sweeper has zeroed reserved_balance but left balance untouched.
+    key = _make_key(balance=1000, reserved=0)
+    integration_session.add(key)
+    await integration_session.commit()
+    await pay_for_request(key, deducted_max_cost, integration_session)
+    key.reserved_balance = 0
+    integration_session.add(key)
+    await integration_session.commit()
+
+    response_data = {
+        "model": "test-model",
+        "usage": {"prompt_tokens": 100, "completion_tokens": 100},
+    }
+
+    with patch(
+        "routstr.auth.calculate_cost",
+        return_value=_cost_data(actual_token_cost),
+    ):
+        await adjust_payment_for_tokens(
+            key, response_data, integration_session, deducted_max_cost, None, None
+        )
+
+    await integration_session.refresh(key)
+
+    assert key.total_spent == actual_token_cost, (
+        f"Request was not billed (total_spent={key.total_spent}) — free response bug"
+    )
+    assert key.balance == 1000 - actual_token_cost, (
+        f"Balance not charged: {key.balance}"
+    )
+    assert key.balance >= 0
+    assert key.reserved_balance == 0
+
+
+@pytest.mark.asyncio
+async def test_free_response_path_closed_end_to_end(
+    integration_session: AsyncSession,
+    patched_db_engine: None,
+) -> None:
+    """A reservation released by the real sweeper must not yield a free response."""
+    from routstr.auth import (
+        adjust_payment_for_tokens,
+        get_reservation_snapshot,
+        pay_for_request,
+    )
+    from routstr.core.db import (
+        ReservationRelease,
+        create_session,
+        release_stale_reservations,
+    )
+
+    deducted_max_cost = 990
+    actual_token_cost = 1000
+    key_hash = f"test_sweep_{uuid.uuid4().hex}"
+
+    async with create_session() as session:
+        session.add(
+            ApiKey(
+                hashed_key=key_hash,
+                balance=1000,
+                reserved_balance=0,
+                total_spent=0,
+                total_requests=0,
+            )
+        )
+        await session.commit()
+
+    # Reserve the request, then backdate reserved_at so the sweeper treats it as
+    # stale (simulates a stream that outlived stale_reservation_timeout_seconds).
+    async with create_session() as session:
+        key = await session.get(ApiKey, key_hash)
+        assert key is not None
+        await pay_for_request(key, deducted_max_cost, session)
+        snapshot = await get_reservation_snapshot(key, session)
+        await session.refresh(key)
+        assert key.reserved_balance == deducted_max_cost
+        key.reserved_at = int(time.time()) - 10_000
+        record = await session.get(ReservationRelease, snapshot.release_id)
+        assert record is not None
+        record.created_at = int(time.time()) - 10_000
+        session.add(key)
+        session.add(record)
+        await session.commit()
+
+    # Sweeper releases the stale reservation without charging.
+    async with create_session() as session:
+        released = await release_stale_reservations(session, max_age_seconds=300)
+        assert released == 1
+
+    async with create_session() as session:
+        key = await session.get(ApiKey, key_hash)
+        assert key is not None
+        assert key.reserved_balance == 0, "Precondition: sweeper zeroed the reservation"
+
+        response_data = {
+            "model": "test-model",
+            "usage": {"prompt_tokens": 100, "completion_tokens": 100},
+        }
+        with patch(
+            "routstr.auth.calculate_cost",
+            return_value=_cost_data(actual_token_cost),
+        ):
+            await adjust_payment_for_tokens(
+                key,
+                response_data,
+                session,
+                deducted_max_cost,
+                reservation_snapshot=snapshot,
+            )
+
+    async with create_session() as session:
+        final = await session.get(ApiKey, key_hash)
+        assert final is not None
+
+    # Stale release is terminal for this reservation. A late finalizer must not
+    # charge aggregate balance that may now belong to a newer request.
+    assert final.total_spent == 0
+    assert final.balance == 1000
+    assert final.balance >= 0
+    assert final.reserved_balance == 0

tests/integration/test_insufficient_balance.py

@@ -241,8 +241,10 @@ async def test_http_402_response_shape_on_insufficient_balance(
     mock_upstream.prepare_headers = MagicMock(return_value={})
 
     with (
-        patch("routstr.proxy.get_model_instance", return_value=mock_model),
-        patch("routstr.proxy.get_provider_for_model", return_value=[mock_upstream]),
+        patch(
+            "routstr.proxy.get_candidates",
+            return_value=[(mock_model, mock_upstream)],
+        ),
         # Patch where it is used (proxy imports it at module level)
         patch(
             "routstr.proxy.get_max_cost_for_model",

tests/integration/test_lightning_invoice_constraints.py

@@ -3,20 +3,23 @@
 Covers two things:
 - The three constraint fields (balance_limit, balance_limit_reset, validity_date)
   are persisted on LightningInvoice and survive a DB round-trip.
-- create_api_key_from_invoice propagates those fields to the created ApiKey,
-  so the constraints are actually enforced when the key is used.
+- The production-path API-key record helper propagates those fields to the
+  created ApiKey, so the constraints are actually enforced when the key is used.
 """
 
 from __future__ import annotations
 
+import asyncio
 import time
-from unittest.mock import AsyncMock, patch
+from unittest.mock import AsyncMock, MagicMock, patch
 
 import pytest
+from sqlalchemy import inspect
+from sqlalchemy.ext.asyncio import AsyncEngine
 from sqlmodel.ext.asyncio.session import AsyncSession
 
 from routstr.core.db import ApiKey, LightningInvoice
-from routstr.lightning import create_api_key_from_invoice
+from routstr.lightning import _create_api_key_record
 
 
 def _make_invoice(**kwargs: object) -> LightningInvoice:
@@ -48,6 +51,7 @@ def mock_wallet_mint() -> object:
 # Persistence
 # ---------------------------------------------------------------------------
 
+
 @pytest.mark.asyncio
 async def test_invoice_persists_balance_limit(
     integration_session: AsyncSession,
@@ -92,6 +96,7 @@ async def test_invoice_persists_validity_date(
 # Propagation to ApiKey
 # ---------------------------------------------------------------------------
 
+
 @pytest.mark.asyncio
 async def test_created_key_receives_balance_limit(
     integration_session: AsyncSession,
@@ -100,7 +105,7 @@ async def test_created_key_receives_balance_limit(
     integration_session.add(invoice)
     await integration_session.flush()
 
-    api_key = await create_api_key_from_invoice(invoice, integration_session)
+    api_key = await _create_api_key_record(invoice, integration_session)
     await integration_session.commit()
 
     stored_key = await integration_session.get(ApiKey, api_key.hashed_key)
@@ -116,7 +121,7 @@ async def test_created_key_receives_balance_limit_reset(
     integration_session.add(invoice)
     await integration_session.flush()
 
-    api_key = await create_api_key_from_invoice(invoice, integration_session)
+    api_key = await _create_api_key_record(invoice, integration_session)
     await integration_session.commit()
 
     stored_key = await integration_session.get(ApiKey, api_key.hashed_key)
@@ -133,7 +138,7 @@ async def test_created_key_receives_validity_date(
     integration_session.add(invoice)
     await integration_session.flush()
 
-    api_key = await create_api_key_from_invoice(invoice, integration_session)
+    api_key = await _create_api_key_record(invoice, integration_session)
     await integration_session.commit()
 
     stored_key = await integration_session.get(ApiKey, api_key.hashed_key)
@@ -141,6 +146,243 @@ async def test_created_key_receives_validity_date(
     assert stored_key.validity_date == expiry
 
 
+@pytest.mark.asyncio
+async def test_payment_check_releases_connection_during_mint_quote(
+    integration_engine: AsyncEngine,
+    patched_db_engine: None,
+) -> None:
+    invoice = _make_invoice(id="inv_slow_quote", status="pending", paid_at=None)
+    async with AsyncSession(integration_engine, expire_on_commit=False) as setup:
+        setup.add(invoice)
+        await setup.commit()
+
+    async with AsyncSession(integration_engine, expire_on_commit=False) as session:
+        stored = await session.get(LightningInvoice, invoice.id)
+        assert stored is not None
+
+        async def quote_status(*args: object, **kwargs: object) -> MagicMock:
+            assert integration_engine.pool.checkedout() == 0  # type: ignore[attr-defined]
+            return MagicMock(paid=False)
+
+        wallet = MagicMock()
+        wallet.get_mint_quote = AsyncMock(side_effect=quote_status)
+        with patch("routstr.lightning.get_wallet", AsyncMock(return_value=wallet)):
+            from routstr.lightning import check_invoice_payment
+
+            await check_invoice_payment(stored, session)
+
+
+@pytest.mark.asyncio
+async def test_concurrent_payment_checks_mint_and_credit_invoice_once(
+    integration_engine: AsyncEngine,
+    patched_db_engine: None,
+) -> None:
+    invoice = _make_invoice(id="inv_concurrent", status="pending", paid_at=None)
+    async with AsyncSession(integration_engine, expire_on_commit=False) as setup:
+        setup.add(invoice)
+        await setup.commit()
+
+    wallet = MagicMock()
+    wallet.get_mint_quote = AsyncMock(return_value=MagicMock(paid=True))
+
+    mint_calls = 0
+
+    async def single_use_mint(*args: object, **kwargs: object) -> list[object]:
+        # Real mints enforce single-use quotes: the second concurrent minter
+        # gets rejected at the mint, mirroring cashu quote semantics.
+        nonlocal mint_calls
+        mint_calls += 1
+        call_number = mint_calls
+        await asyncio.sleep(0.05)
+        if call_number > 1:
+            raise Exception("quote already issued")
+        return []
+
+    wallet.mint = AsyncMock(side_effect=single_use_mint)
+
+    async with (
+        AsyncSession(integration_engine, expire_on_commit=False) as first,
+        AsyncSession(integration_engine, expire_on_commit=False) as second,
+    ):
+        first_invoice = await first.get(LightningInvoice, invoice.id)
+        second_invoice = await second.get(LightningInvoice, invoice.id)
+        assert first_invoice is not None
+        assert second_invoice is not None
+
+        with patch("routstr.lightning.get_wallet", AsyncMock(return_value=wallet)):
+            from routstr.lightning import check_invoice_payment
+
+            await asyncio.gather(
+                check_invoice_payment(first_invoice, first),
+                check_invoice_payment(second_invoice, second),
+            )
+
+    async with AsyncSession(integration_engine, expire_on_commit=False) as verify:
+        stored_invoice = await verify.get(LightningInvoice, invoice.id)
+        assert stored_invoice is not None
+        assert stored_invoice.status == "paid"
+        assert stored_invoice.api_key_hash is not None
+        stored_key = await verify.get(ApiKey, stored_invoice.api_key_hash)
+        assert stored_key is not None
+        assert stored_key.balance == invoice.amount_sats * 1000
+
+
+@pytest.mark.asyncio
+async def test_failed_mint_keeps_invoice_pending_for_retry(
+    integration_engine: AsyncEngine,
+    patched_db_engine: None,
+) -> None:
+    invoice = _make_invoice(id="inv_mint_failure", status="pending", paid_at=None)
+    async with AsyncSession(integration_engine, expire_on_commit=False) as setup:
+        setup.add(invoice)
+        await setup.commit()
+
+    wallet = MagicMock()
+    wallet.get_mint_quote = AsyncMock(return_value=MagicMock(paid=True))
+    wallet.mint = AsyncMock(side_effect=TimeoutError("mint unavailable"))
+    async with AsyncSession(integration_engine, expire_on_commit=False) as session:
+        stored = await session.get(LightningInvoice, invoice.id)
+        assert stored is not None
+        with patch("routstr.lightning.get_wallet", AsyncMock(return_value=wallet)):
+            from routstr.lightning import check_invoice_payment
+
+            await check_invoice_payment(stored, session)
+
+    async with AsyncSession(integration_engine, expire_on_commit=False) as verify:
+        stored = await verify.get(LightningInvoice, invoice.id)
+        assert stored is not None
+        assert stored.status == "pending"
+
+
+@pytest.mark.asyncio
+async def test_unpaid_topup_does_not_query_target_key(
+    integration_engine: AsyncEngine,
+    patched_db_engine: None,
+) -> None:
+    invoice = _make_invoice(
+        id="inv_unpaid_topup",
+        status="pending",
+        paid_at=None,
+        purpose="topup",
+        api_key_hash="target-key",
+    )
+    async with AsyncSession(integration_engine, expire_on_commit=False) as setup:
+        setup.add(invoice)
+        await setup.commit()
+
+    wallet = MagicMock()
+    wallet.get_mint_quote = AsyncMock(return_value=MagicMock(paid=False))
+    create_session = MagicMock(side_effect=RuntimeError("target lookup should not run"))
+    async with AsyncSession(integration_engine, expire_on_commit=False) as session:
+        stored = await session.get(LightningInvoice, invoice.id)
+        assert stored is not None
+        with (
+            patch("routstr.lightning.get_wallet", AsyncMock(return_value=wallet)),
+            patch("routstr.lightning.create_session", create_session),
+        ):
+            from routstr.lightning import check_invoice_payment
+
+            await check_invoice_payment(stored, session)
+
+    wallet.get_mint_quote.assert_awaited_once_with(invoice.payment_hash)
+    create_session.assert_not_called()
+
+
+@pytest.mark.asyncio
+async def test_missing_topup_target_is_rejected_before_mint(
+    integration_engine: AsyncEngine,
+    patched_db_engine: None,
+) -> None:
+    invoice = _make_invoice(
+        id="inv_missing_topup_target",
+        status="pending",
+        paid_at=None,
+        purpose="topup",
+        api_key_hash="pruned-key",
+        expires_at=int(time.time()) - 1,
+    )
+    async with AsyncSession(integration_engine, expire_on_commit=False) as setup:
+        setup.add(invoice)
+        await setup.commit()
+
+    wallet = MagicMock()
+    wallet.get_mint_quote = AsyncMock(return_value=MagicMock(paid=True))
+    wallet.mint = AsyncMock()
+    async with AsyncSession(integration_engine, expire_on_commit=False) as session:
+        stored = await session.get(LightningInvoice, invoice.id)
+        assert stored is not None
+        with (
+            patch("routstr.lightning.get_wallet", AsyncMock(return_value=wallet)),
+            patch("routstr.lightning.logger.critical") as critical,
+        ):
+            from routstr.lightning import get_invoice_status
+
+            response = await get_invoice_status(invoice.id, session)
+
+        assert response.status == "reconciliation_required"
+        assert stored.status == "reconciliation_required"
+        assert stored not in session.dirty
+        critical.assert_called_once()
+
+    wallet.mint.assert_not_awaited()
+    async with AsyncSession(integration_engine) as verify:
+        stored = await verify.get(LightningInvoice, invoice.id)
+        assert stored is not None
+        assert stored.status == "reconciliation_required"
+
+
+@pytest.mark.asyncio
+async def test_post_mint_db_failure_keeps_invoice_pending_for_reconciliation(
+    integration_engine: AsyncEngine,
+    patched_db_engine: None,
+) -> None:
+    invoice = _make_invoice(id="inv_finalize_failure", status="pending", paid_at=None)
+    sibling = _make_invoice(
+        id="inv_finalize_failure_sibling",
+        bolt11="lnbc1000n1sibling",
+        payment_hash="cafebabe" * 8,
+        status="pending",
+        paid_at=None,
+    )
+    async with AsyncSession(integration_engine, expire_on_commit=False) as setup:
+        setup.add_all([invoice, sibling])
+        await setup.commit()
+
+    wallet = MagicMock()
+    wallet.get_mint_quote = AsyncMock(return_value=MagicMock(paid=True))
+    wallet.mint = AsyncMock(return_value=[])
+    async with AsyncSession(integration_engine, expire_on_commit=False) as session:
+        stored = await session.get(LightningInvoice, invoice.id)
+        stored_sibling = await session.get(LightningInvoice, sibling.id)
+        assert stored is not None
+        assert stored_sibling is not None
+        with (
+            patch("routstr.lightning.get_wallet", AsyncMock(return_value=wallet)),
+            patch(
+                "routstr.lightning._create_api_key_record",
+                AsyncMock(side_effect=RuntimeError("database unavailable")),
+            ),
+        ):
+            from routstr.lightning import check_invoice_payment
+
+            await check_invoice_payment(stored, session)
+
+        stored_state = inspect(stored)
+        sibling_state = inspect(stored_sibling)
+        assert stored_state is not None
+        assert sibling_state is not None
+        assert stored_state.expired is False
+        assert sibling_state.expired is False
+        assert stored.status == "pending"
+        assert stored_sibling.id == sibling.id
+
+    assert wallet.mint.await_count == 1
+    async with AsyncSession(integration_engine, expire_on_commit=False) as verify:
+        stored = await verify.get(LightningInvoice, invoice.id)
+        assert stored is not None
+        assert stored.status == "pending"
+
+
 @pytest.mark.asyncio
 async def test_created_key_without_constraints_has_none_fields(
     integration_session: AsyncSession,
@@ -149,7 +391,7 @@ async def test_created_key_without_constraints_has_none_fields(
     integration_session.add(invoice)
     await integration_session.flush()
 
-    api_key = await create_api_key_from_invoice(invoice, integration_session)
+    api_key = await _create_api_key_record(invoice, integration_session)
     await integration_session.commit()
 
     stored_key = await integration_session.get(ApiKey, api_key.hashed_key)
@@ -157,3 +399,84 @@ async def test_created_key_without_constraints_has_none_fields(
     assert stored_key.balance_limit is None
     assert stored_key.balance_limit_reset is None
     assert stored_key.validity_date is None
+
+
+@pytest.mark.asyncio
+async def test_db_guard_credits_once_when_both_mints_succeed(
+    integration_engine: AsyncEngine,
+    patched_db_engine: None,
+) -> None:
+    """Even if the mint fails to enforce single-use quotes and both racers
+    mint successfully, the conditional status update must credit exactly once."""
+    key = ApiKey(hashed_key="race-key", balance=1_000)
+    invoice = _make_invoice(
+        id="inv_db_guard",
+        status="pending",
+        paid_at=None,
+        purpose="topup",
+        api_key_hash="race-key",
+    )
+    sibling = _make_invoice(
+        id="inv_db_guard_sibling",
+        bolt11="lnbc1000n1race-sibling",
+        payment_hash="01234567" * 8,
+        status="pending",
+        paid_at=None,
+    )
+    async with AsyncSession(integration_engine, expire_on_commit=False) as setup:
+        setup.add_all([key, invoice, sibling])
+        await setup.commit()
+
+    wallet = MagicMock()
+    wallet.get_mint_quote = AsyncMock(return_value=MagicMock(paid=True))
+
+    async def always_succeeding_mint(*args: object, **kwargs: object) -> list[object]:
+        await asyncio.sleep(0.05)
+        return []
+
+    wallet.mint = AsyncMock(side_effect=always_succeeding_mint)
+
+    async with (
+        AsyncSession(integration_engine, expire_on_commit=False) as first,
+        AsyncSession(integration_engine, expire_on_commit=False) as second,
+    ):
+        first_invoice = await first.get(LightningInvoice, invoice.id)
+        first_sibling = await first.get(LightningInvoice, sibling.id)
+        second_invoice = await second.get(LightningInvoice, invoice.id)
+        assert first_invoice is not None
+        assert first_sibling is not None
+        assert second_invoice is not None
+
+        with patch("routstr.lightning.get_wallet", AsyncMock(return_value=wallet)):
+            from routstr.lightning import check_invoice_payment
+
+            await asyncio.gather(
+                check_invoice_payment(first_invoice, first),
+                check_invoice_payment(second_invoice, second),
+            )
+
+        first_state = inspect(first_invoice)
+        sibling_state = inspect(first_sibling)
+        second_state = inspect(second_invoice)
+        assert first_state is not None
+        assert sibling_state is not None
+        assert second_state is not None
+        assert first_state.expired is False
+        assert sibling_state.expired is False
+        assert second_state.expired is False
+        assert first_invoice.id == invoice.id
+        assert first_sibling.id == sibling.id
+        assert second_invoice.id == invoice.id
+        assert first_invoice.status == "paid"
+        assert second_invoice.status == "paid"
+        assert first_invoice not in first.dirty
+        assert second_invoice not in second.dirty
+
+    assert wallet.mint.await_count == 2
+    async with AsyncSession(integration_engine, expire_on_commit=False) as verify:
+        stored_invoice = await verify.get(LightningInvoice, invoice.id)
+        assert stored_invoice is not None
+        assert stored_invoice.status == "paid"
+        stored_key = await verify.get(ApiKey, "race-key")
+        assert stored_key is not None
+        assert stored_key.balance == 1_000 + invoice.amount_sats * 1000

tests/integration/test_periodic_payout_safety.py

@@ -0,0 +1,174 @@
+"""Money-safety regression coverage for automatic wallet payouts."""
+
+from __future__ import annotations
+
+import asyncio
+from collections.abc import Callable, Coroutine
+from typing import Any
+from unittest.mock import AsyncMock, MagicMock, patch
+
+import pytest
+from sqlalchemy.ext.asyncio import AsyncEngine
+from sqlmodel.ext.asyncio.session import AsyncSession
+
+from routstr.core import db
+from routstr.core.db import ApiKey
+from routstr.core.settings import settings
+from routstr.wallet import credit_balance, periodic_payout
+
+PRIMARY_MINT = "http://primary:3338"
+REFUND_MINT = "http://refund:3338"
+PAYOUT_INTERVAL = 987
+
+
+class _LoopBreak(Exception):
+    """Stop the otherwise-infinite payout loop after one cycle."""
+
+
+def _one_payout_cycle() -> Callable[[float], Coroutine[Any, Any, None]]:
+    intervals_seen = 0
+
+    async def sleep(seconds: float) -> None:
+        nonlocal intervals_seen
+        if seconds == PAYOUT_INTERVAL:
+            intervals_seen += 1
+            if intervals_seen == 2:
+                raise _LoopBreak()
+
+    return sleep
+
+
+@pytest.mark.asyncio
+async def test_cross_mint_liability_is_not_paid_as_owner_profit(
+    integration_engine: AsyncEngine,
+    patched_db_engine: None,
+) -> None:
+    """Refund preferences must not make primary-mint customer funds payable."""
+    async with AsyncSession(integration_engine, expire_on_commit=False) as setup:
+        setup.add(
+            ApiKey(
+                hashed_key="cross-mint-key",
+                balance=50_000,
+                refund_mint_url=REFUND_MINT,
+                refund_currency="sat",
+            )
+        )
+        await setup.commit()
+
+    primary_proof = MagicMock(amount=50)
+    raw_send = AsyncMock(return_value=50)
+
+    def proofs_for_mint(
+        _wallet: object, mint_url: str, unit: str, **_kwargs: object
+    ) -> list[MagicMock]:
+        if mint_url == PRIMARY_MINT and unit == "sat":
+            return [primary_proof]
+        return []
+
+    with (
+        patch.object(settings, "cashu_mints", [REFUND_MINT]),
+        patch.object(settings, "primary_mint", PRIMARY_MINT),
+        patch.object(settings, "receive_ln_address", "owner@ln.test"),
+        patch.object(settings, "payout_interval_seconds", PAYOUT_INTERVAL),
+        patch.object(settings, "min_payout_sat", 10),
+        patch("routstr.wallet.asyncio.sleep", _one_payout_cycle()),
+        patch("routstr.wallet.get_wallet", AsyncMock(return_value=MagicMock())),
+        patch(
+            "routstr.wallet.get_proofs_per_mint_and_unit",
+            MagicMock(side_effect=proofs_for_mint),
+        ),
+        patch(
+            "routstr.wallet.slow_filter_spend_proofs",
+            AsyncMock(side_effect=lambda proofs, _wallet: proofs),
+        ),
+        patch("routstr.wallet.raw_send_to_lnurl", raw_send),
+    ):
+        with pytest.raises(_LoopBreak):
+            await periodic_payout()
+
+    raw_send.assert_not_awaited()
+
+
+@pytest.mark.asyncio
+async def test_payout_does_not_send_proofs_whose_liability_commit_is_in_flight(
+    integration_engine: AsyncEngine,
+    patched_db_engine: None,
+) -> None:
+    """Proof visibility before liability commit must not expose customer funds."""
+    key = ApiKey(
+        hashed_key="in-flight-topup-key",
+        balance=0,
+        refund_mint_url=PRIMARY_MINT,
+        refund_currency="sat",
+    )
+    async with AsyncSession(integration_engine, expire_on_commit=False) as setup:
+        setup.add(key)
+        await setup.commit()
+
+    proofs: list[MagicMock] = []
+    proof_visible = asyncio.Event()
+    finish_redemption = asyncio.Event()
+    liability_read = asyncio.Event()
+
+    async def redeem_token(token: str) -> tuple[int, str, str]:
+        proofs.append(MagicMock(amount=200))
+        proof_visible.set()
+        await finish_redemption.wait()
+        return 200, "sat", PRIMARY_MINT
+
+    real_total_liability = db.total_user_liability
+
+    async def read_liability(_session: AsyncSession) -> int:
+        async with db.create_session() as snapshot_session:
+            value = await real_total_liability(snapshot_session)
+        liability_read.set()
+        return value
+
+    raw_send = AsyncMock(return_value=200)
+
+    with (
+        patch.object(settings, "cashu_mints", []),
+        patch.object(settings, "primary_mint", PRIMARY_MINT),
+        patch.object(settings, "receive_ln_address", "owner@ln.test"),
+        patch.object(settings, "payout_interval_seconds", PAYOUT_INTERVAL),
+        patch.object(settings, "min_payout_sat", 10),
+        patch("routstr.wallet.asyncio.sleep", _one_payout_cycle()),
+        patch("routstr.wallet.recieve_token", AsyncMock(side_effect=redeem_token)),
+        patch("routstr.wallet.get_wallet", AsyncMock(return_value=MagicMock())),
+        patch(
+            "routstr.wallet.get_proofs_per_mint_and_unit",
+            MagicMock(side_effect=lambda *_args, **_kwargs: list(proofs)),
+        ),
+        patch(
+            "routstr.wallet.slow_filter_spend_proofs",
+            AsyncMock(side_effect=lambda visible, _wallet: visible),
+        ),
+        patch(
+            "routstr.wallet.db.total_user_liability",
+            AsyncMock(side_effect=read_liability),
+        ),
+        patch("routstr.wallet.raw_send_to_lnurl", raw_send),
+    ):
+        async with AsyncSession(integration_engine, expire_on_commit=False) as credit_session:
+            stored_key = await credit_session.get(ApiKey, key.hashed_key)
+            assert stored_key is not None
+            credit_task = asyncio.create_task(
+                credit_balance("cashu-token", stored_key, credit_session)
+            )
+            await asyncio.wait_for(proof_visible.wait(), timeout=2)
+
+            payout_task = asyncio.create_task(periodic_payout())
+            try:
+                await asyncio.wait_for(liability_read.wait(), timeout=0.1)
+                liability_was_read_while_crediting = True
+            except TimeoutError:
+                liability_was_read_while_crediting = False
+
+            finish_redemption.set()
+            await asyncio.wait_for(credit_task, timeout=2)
+
+            with pytest.raises(_LoopBreak):
+                await asyncio.wait_for(payout_task, timeout=2)
+
+    assert liability_was_read_while_crediting is False
+    raw_send.assert_not_awaited()

tests/integration/test_provider_self_lookup.py

@@ -0,0 +1,128 @@
+"""A live upstream provider resolves its OWN database row by stable identity
+(its primary key), not by its mutable/secret ``api_key``.
+
+Today ``from_db_row`` drops ``provider_row.id`` and the two self-referential
+paths — PPQ.AI's insufficient-balance self-disable and the base
+``refresh_models_cache`` — re-find their own row with
+``WHERE base_url == self.base_url AND api_key == self.api_key``. That uses a
+rotatable secret as a self-handle: the moment the row's key changes underneath a
+live object (a rotation racing an in-flight request), the object can no longer
+find itself. These tests pin the invariant that a provider looks itself up by
+identity, so the lookup survives a key change (and, later, key encryption).
+"""
+
+from unittest.mock import AsyncMock, patch
+
+import pytest
+
+from routstr.core.db import UpstreamProviderRow
+from routstr.upstream.ppqai import PPQAIUpstreamProvider
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_provider_object_carries_its_persistent_identity(
+    integration_session: object,
+    patched_db_engine: None,
+) -> None:
+    """``from_db_row`` gives the in-memory object its row's identity (``db_id``)."""
+    row = UpstreamProviderRow(
+        provider_type="ppqai",
+        base_url="https://api.ppq.ai",
+        api_key="sk-original",
+        enabled=True,
+        provider_fee=1.0,
+    )
+    integration_session.add(row)  # type: ignore[attr-defined]
+    await integration_session.commit()  # type: ignore[attr-defined]
+    await integration_session.refresh(row)  # type: ignore[attr-defined]
+
+    provider = PPQAIUpstreamProvider.from_db_row(row)
+    assert provider is not None
+
+    assert provider.db_id == row.id
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_self_disable_targets_own_row_after_key_rotation(
+    integration_session: object,
+    patched_db_engine: None,
+) -> None:
+    """PPQ.AI self-disable must disable *its* row even after the key rotated.
+
+    RED (current): the object holds the pre-rotation key, so the
+    ``(base_url, api_key)`` lookup misses the row → the provider is never
+    disabled. GREEN: lookup by ``id`` finds it and disables it.
+    """
+    row = UpstreamProviderRow(
+        provider_type="ppqai",
+        base_url="https://api.ppq.ai",
+        api_key="sk-original",
+        enabled=True,
+        provider_fee=1.0,
+    )
+    integration_session.add(row)  # type: ignore[attr-defined]
+    await integration_session.commit()  # type: ignore[attr-defined]
+    await integration_session.refresh(row)  # type: ignore[attr-defined]
+
+    provider = PPQAIUpstreamProvider.from_db_row(row)  # captures sk-original
+    assert provider is not None
+
+    # Key is rotated in the DB while `provider` is still live.
+    row.api_key = "sk-rotated"
+    integration_session.add(row)  # type: ignore[attr-defined]
+    await integration_session.commit()  # type: ignore[attr-defined]
+
+    with patch("routstr.proxy.reinitialize_upstreams", new=AsyncMock()):
+        await provider.on_upstream_error_redirect(402, "Insufficient balance")
+
+    await integration_session.refresh(row)  # type: ignore[attr-defined]
+    assert row.enabled is False
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_refresh_models_cache_finds_own_row_after_key_rotation(
+    integration_session: object,
+    patched_db_engine: None,
+) -> None:
+    """``refresh_models_cache`` must resolve its own row after a key rotation.
+
+    ``refresh_models_cache`` swallows every exception (it only logs), so the
+    observable proof it found its row is that it reaches ``list_models`` — which
+    is called with the row's ``id`` only *after* the row is resolved. RED
+    (current): the stale-key ``(base_url, api_key)`` lookup returns nothing, the
+    method raises ``404`` internally and returns before ``list_models`` is ever
+    called. GREEN: lookup by ``id`` finds the row and ``list_models`` runs for
+    that ``id``.
+    """
+    row = UpstreamProviderRow(
+        provider_type="ppqai",
+        base_url="https://api.ppq.ai",
+        api_key="sk-original",
+        enabled=True,
+        provider_fee=1.0,
+    )
+    integration_session.add(row)  # type: ignore[attr-defined]
+    await integration_session.commit()  # type: ignore[attr-defined]
+    await integration_session.refresh(row)  # type: ignore[attr-defined]
+    row_id = row.id
+
+    provider = PPQAIUpstreamProvider.from_db_row(row)  # captures sk-original
+    assert provider is not None
+
+    row.api_key = "sk-rotated"
+    integration_session.add(row)  # type: ignore[attr-defined]
+    await integration_session.commit()  # type: ignore[attr-defined]
+
+    list_models_mock = AsyncMock(return_value=[])
+    with (
+        patch.object(provider, "fetch_models", new=AsyncMock(return_value=[])),
+        patch("routstr.upstream.base.list_models", new=list_models_mock),
+    ):
+        await provider.refresh_models_cache()
+
+    list_models_mock.assert_awaited_once()
+    assert list_models_mock.await_args is not None
+    assert list_models_mock.await_args.kwargs["upstream_id"] == row_id

tests/integration/test_proxy_session_lifecycle.py

@@ -0,0 +1,65 @@
+"""Integration coverage for proxy database-session lifetime."""
+
+from __future__ import annotations
+
+import json
+from unittest.mock import AsyncMock, MagicMock, patch
+
+import pytest
+from fastapi import Response
+from sqlalchemy.ext.asyncio import AsyncEngine
+from sqlmodel.ext.asyncio.session import AsyncSession
+
+from routstr import proxy as proxy_module
+from routstr.core.db import ApiKey
+
+
+@pytest.mark.asyncio
+async def test_authenticated_proxy_releases_db_connection_before_upstream_headers(
+    integration_engine: AsyncEngine,
+    integration_session: AsyncSession,
+    patched_db_engine: None,
+) -> None:
+    """Slow upstream header waits must not retain a checked-out DB connection."""
+    key = ApiKey(
+        hashed_key="proxy-pool-key",
+        balance=1_000_000,
+        refund_mint_url="http://primary:3338",
+        refund_currency="sat",
+    )
+    integration_session.add(key)
+    await integration_session.commit()
+
+    request = MagicMock()
+    request.method = "POST"
+    request.headers = {"authorization": "Bearer test-key"}
+    request.body = AsyncMock(return_value=json.dumps({"model": "test-model"}).encode())
+    request.url.path = "/v1/chat/completions"
+    request.state.request_id = "pool-hold-regression"
+
+    model = MagicMock()
+    upstream = MagicMock()
+    upstream.provider_type = "test"
+    upstream.prepare_headers.return_value = {}
+
+    async def wait_for_headers(*args: object, **kwargs: object) -> Response:
+        assert integration_engine.pool.checkedout() == 0  # type: ignore[attr-defined]
+        return Response(status_code=200)
+
+    upstream.forward_request = AsyncMock(side_effect=wait_for_headers)
+
+    with (
+        patch("routstr.proxy.get_candidates", return_value=[(model, upstream)]),
+        patch("routstr.proxy.get_max_cost_for_model", AsyncMock(return_value=100)),
+        patch(
+            "routstr.proxy.calculate_discounted_max_cost",
+            AsyncMock(return_value=100),
+        ),
+        patch("routstr.proxy.check_token_balance"),
+        patch("routstr.proxy.get_bearer_token_key", AsyncMock(return_value=key)),
+    ):
+        response = await proxy_module._proxy(
+            request, "v1/chat/completions", integration_session
+        )
+
+    assert response.status_code == 200

tests/integration/test_prune_dead_api_keys.py

@@ -0,0 +1,283 @@
+"""
+Tests for prune_dead_api_keys — the janitor that removes provably-dead 0/0/0
+API keys (funded keys fully refunded/expired without ever being used, plus bare
+orphans), while protecting keys that are still meaningful.
+"""
+
+import time
+import uuid
+from collections.abc import Awaitable, Callable
+from typing import Any, cast
+
+import pytest
+from sqlalchemy.sql.dml import Update
+from sqlmodel import col, update
+
+from routstr.core.db import (
+    ApiKey,
+    CashuTransaction,
+    LightningInvoice,
+    create_session,
+    prune_dead_api_keys,
+)
+
+OLD = 100  # min_age_seconds used by the tests
+NOW = int(time.time())
+LONG_AGO = NOW - 10_000  # well past the grace period
+
+
+async def _exists(key_hash: str) -> bool:
+    async with create_session() as session:
+        return (await session.get(ApiKey, key_hash)) is not None
+
+
+def _dead_key(created_at: int | None) -> ApiKey:
+    return ApiKey(
+        hashed_key=f"dead_{uuid.uuid4().hex}",
+        balance=0,
+        reserved_balance=0,
+        total_spent=0,
+        total_requests=0,
+        created_at=created_at,
+    )
+
+
+@pytest.mark.asyncio
+async def test_prunes_old_refunded_zero_key(patched_db_engine: None) -> None:
+    """A funded-then-refunded key (0/0/0, NULL parent, old) is pruned."""
+    key = _dead_key(LONG_AGO)
+    async with create_session() as session:
+        session.add(key)
+        await session.commit()
+
+    async with create_session() as session:
+        pruned = await prune_dead_api_keys(session, OLD)
+
+    assert pruned == 1
+    assert not await _exists(key.hashed_key)
+
+
+@pytest.mark.asyncio
+async def test_grace_period_protects_fresh_key(patched_db_engine: None) -> None:
+    """A dead-looking but recently created key is protected by the grace period."""
+    key = _dead_key(int(time.time()))
+    async with create_session() as session:
+        session.add(key)
+        await session.commit()
+
+    async with create_session() as session:
+        pruned = await prune_dead_api_keys(session, OLD)
+
+    assert pruned == 0
+    assert await _exists(key.hashed_key)
+
+
+@pytest.mark.asyncio
+async def test_used_key_never_pruned(patched_db_engine: None) -> None:
+    """Keys with any spend/requests or live balance are never pruned."""
+    spent = _dead_key(LONG_AGO)
+    spent.total_spent = 1
+    requested = _dead_key(LONG_AGO)
+    requested.total_requests = 1
+    funded = _dead_key(LONG_AGO)
+    funded.balance = 1000
+    reserved = _dead_key(LONG_AGO)
+    reserved.reserved_balance = 500
+
+    async with create_session() as session:
+        for k in (spent, requested, funded, reserved):
+            session.add(k)
+        await session.commit()
+
+    async with create_session() as session:
+        pruned = await prune_dead_api_keys(session, OLD)
+
+    assert pruned == 0
+    for k in (spent, requested, funded, reserved):
+        assert await _exists(k.hashed_key)
+
+
+@pytest.mark.asyncio
+async def test_parent_and_child_keys_are_not_pruned(
+    patched_db_engine: None,
+) -> None:
+    """Pruning must not orphan child keys or delete valid children."""
+    parent = _dead_key(LONG_AGO)
+    child = ApiKey(
+        hashed_key=f"child_{uuid.uuid4().hex}",
+        balance=0,
+        reserved_balance=0,
+        total_spent=0,
+        total_requests=0,
+        created_at=LONG_AGO,
+        parent_key_hash=parent.hashed_key,
+    )
+    async with create_session() as session:
+        session.add(parent)
+        session.add(child)
+        await session.commit()
+
+    async with create_session() as session:
+        pruned = await prune_dead_api_keys(session, OLD)
+
+    assert pruned == 0
+    assert await _exists(parent.hashed_key)
+    assert await _exists(child.hashed_key)
+
+
+@pytest.mark.asyncio
+async def test_pending_invoice_protects_key(patched_db_engine: None) -> None:
+    """A key referenced by a pending topup invoice is never pruned mid-topup."""
+    key = _dead_key(LONG_AGO)
+    invoice = LightningInvoice(
+        id=f"inv_{uuid.uuid4().hex}",
+        bolt11=f"lnbc_{uuid.uuid4().hex}",
+        amount_sats=10,
+        description="topup",
+        payment_hash=uuid.uuid4().hex,
+        status="pending",
+        api_key_hash=key.hashed_key,
+        purpose="topup",
+        expires_at=NOW + 10_000,
+    )
+    async with create_session() as session:
+        session.add(key)
+        session.add(invoice)
+        await session.commit()
+
+    async with create_session() as session:
+        pruned = await prune_dead_api_keys(session, OLD)
+
+    assert pruned == 0
+    assert await _exists(key.hashed_key)
+
+
+@pytest.mark.asyncio
+async def test_paid_invoice_does_not_protect_key(patched_db_engine: None) -> None:
+    """A settled (non-pending) invoice does not keep a dead key alive."""
+    key = _dead_key(LONG_AGO)
+    invoice = LightningInvoice(
+        id=f"inv_{uuid.uuid4().hex}",
+        bolt11=f"lnbc_{uuid.uuid4().hex}",
+        amount_sats=10,
+        description="topup",
+        payment_hash=uuid.uuid4().hex,
+        status="paid",
+        api_key_hash=key.hashed_key,
+        purpose="topup",
+        expires_at=NOW - 1,
+    )
+    async with create_session() as session:
+        session.add(key)
+        session.add(invoice)
+        await session.commit()
+
+    async with create_session() as session:
+        pruned = await prune_dead_api_keys(session, OLD)
+
+    assert pruned == 1
+    assert not await _exists(key.hashed_key)
+
+
+@pytest.mark.asyncio
+async def test_key_that_becomes_meaningful_during_prune_survives(
+    patched_db_engine: None, monkeypatch: pytest.MonkeyPatch
+) -> None:
+    """Revalidate before unlink/delete so a late top-up cannot be pruned."""
+    key = _dead_key(LONG_AGO)
+    txn = CashuTransaction(
+        id=uuid.uuid4().hex,
+        token="cashuABC",
+        amount=21,
+        unit="sat",
+        type="in",
+        source="apikey",
+        api_key_hashed_key=key.hashed_key,
+    )
+    async with create_session() as session:
+        session.add(key)
+        session.add(txn)
+        await session.commit()
+
+    async with create_session() as session:
+        original_exec = cast(Callable[..., Awaitable[Any]], session.exec)
+        topped_up = False
+
+        async def exec_with_late_topup(
+            statement: Any, *args: Any, **kwargs: Any
+        ) -> Any:
+            nonlocal topped_up
+            if (
+                not topped_up
+                and isinstance(statement, Update)
+                and getattr(statement.table, "name", None) == "cashu_transactions"
+            ):
+                topped_up = True
+                async with create_session() as topup_session:
+                    await topup_session.exec(  # type: ignore[call-overload]
+                        update(ApiKey)
+                        .where(col(ApiKey.hashed_key) == key.hashed_key)
+                        .values(balance=42)
+                    )
+                    await topup_session.commit()
+            return await original_exec(statement, *args, **kwargs)
+
+        monkeypatch.setattr(session, "exec", exec_with_late_topup)
+        pruned = await prune_dead_api_keys(session, OLD)
+
+    assert pruned == 0
+    assert await _exists(key.hashed_key)
+    async with create_session() as session:
+        surviving = await session.get(CashuTransaction, txn.id)
+        assert surviving is not None
+        assert surviving.api_key_hashed_key == key.hashed_key
+
+
+@pytest.mark.asyncio
+async def test_transaction_audit_trail_preserved(patched_db_engine: None) -> None:
+    """Pruning a refunded key keeps its cashu_transactions, unlinked from the key."""
+    key = _dead_key(LONG_AGO)
+    txn = CashuTransaction(
+        id=uuid.uuid4().hex,
+        token="cashuABC",
+        amount=21,
+        unit="sat",
+        type="in",
+        source="apikey",
+        api_key_hashed_key=key.hashed_key,
+    )
+    async with create_session() as session:
+        session.add(key)
+        session.add(txn)
+        await session.commit()
+
+    async with create_session() as session:
+        pruned = await prune_dead_api_keys(session, OLD)
+
+    assert pruned == 1
+    assert not await _exists(key.hashed_key)
+
+    async with create_session() as session:
+        surviving = await session.get(CashuTransaction, txn.id)
+        assert surviving is not None, "Financial audit row must survive key deletion"
+        assert surviving.api_key_hashed_key is None, "Link must be nulled, not dangling"
+        assert surviving.amount == 21
+
+
+@pytest.mark.asyncio
+async def test_periodic_prune_disabled_returns_immediately(
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    """Non-positive intervals disable the janitor."""
+    from unittest.mock import AsyncMock
+
+    from routstr import auth
+    from routstr.core.settings import settings
+
+    monkeypatch.setattr(settings, "dead_key_prune_interval_seconds", 0)
+    sleep_mock = AsyncMock()
+    monkeypatch.setattr(auth.asyncio, "sleep", sleep_mock)
+
+    await auth.periodic_dead_key_prune()
+
+    sleep_mock.assert_not_called()

tests/integration/test_reservation_lifecycle.py

@@ -120,7 +120,9 @@ async def test_finalise_releases_reservation_and_charges_balance(
     response_data = {"model": "test-model", "usage": {"prompt_tokens": 50, "completion_tokens": 50}}
 
     with patch("routstr.auth.calculate_cost", return_value=cost_data):
-        await adjust_payment_for_tokens(key, response_data, integration_session, cost)
+        await adjust_payment_for_tokens(
+            key, response_data, integration_session, cost, None, None
+        )
 
     await integration_session.refresh(key)
 

tests/integration/test_reserved_balance_negative.py

@@ -141,7 +141,7 @@ async def test_revert_with_zero_reserved_balance_is_noop(
     Previously this would drive reserved_balance negative. With the floor guard,
     it should return False and leave reserved_balance at 0.
     """
-    from routstr.auth import revert_pay_for_request
+    from routstr.auth import pay_for_request, revert_pay_for_request
 
     unique_key = f"test_revert_key_{uuid.uuid4().hex[:8]}"
     test_key = ApiKey(
@@ -151,8 +151,12 @@ async def test_revert_with_zero_reserved_balance_is_noop(
     )
     integration_session.add(test_key)
     await integration_session.commit()
+    await pay_for_request(test_key, 100, integration_session)
+    test_key.reserved_balance = 0
+    integration_session.add(test_key)
+    await integration_session.commit()
 
-    # Try to revert more than available — should be a no-op
+    # A stale cleanup already released the aggregate reservation.
     result = await revert_pay_for_request(test_key, integration_session, 100)
 
     await integration_session.refresh(test_key)
@@ -161,8 +165,8 @@ async def test_revert_with_zero_reserved_balance_is_noop(
     assert test_key.reserved_balance == 0, (
         f"Reserved balance should remain 0, got: {test_key.reserved_balance}"
     )
-    assert test_key.total_requests == 0, (
-        f"Total requests should remain 0, got: {test_key.total_requests}"
+    assert test_key.total_requests == 1, (
+        f"Total requests should remain 1, got: {test_key.total_requests}"
     )
 
 
@@ -171,17 +175,18 @@ async def test_revert_with_sufficient_reserved_balance_succeeds(
     integration_session: AsyncSession,
 ) -> None:
     """Test that revert_pay_for_request works correctly when there is enough reserved balance."""
-    from routstr.auth import revert_pay_for_request
+    from routstr.auth import pay_for_request, revert_pay_for_request
 
     unique_key = f"test_revert_ok_{uuid.uuid4().hex[:8]}"
     test_key = ApiKey(
         hashed_key=unique_key,
         balance=5000,
-        reserved_balance=500,
-        total_requests=3,
+        reserved_balance=0,
+        total_requests=2,
     )
     integration_session.add(test_key)
     await integration_session.commit()
+    await pay_for_request(test_key, 500, integration_session)
 
     result = await revert_pay_for_request(test_key, integration_session, 500)
 
@@ -202,17 +207,21 @@ async def test_revert_partial_reserved_balance_is_noop(
     integration_session: AsyncSession,
 ) -> None:
     """Test that reverting more than the current reserved_balance is a no-op."""
-    from routstr.auth import revert_pay_for_request
+    from routstr.auth import pay_for_request, revert_pay_for_request
 
     unique_key = f"test_revert_partial_{uuid.uuid4().hex[:8]}"
     test_key = ApiKey(
         hashed_key=unique_key,
         balance=5000,
-        reserved_balance=50,
-        total_requests=1,
+        reserved_balance=0,
+        total_requests=0,
     )
     integration_session.add(test_key)
     await integration_session.commit()
+    await pay_for_request(test_key, 500, integration_session)
+    test_key.reserved_balance = 50
+    integration_session.add(test_key)
+    await integration_session.commit()
 
     # Try to revert 500 when only 50 is reserved — should be no-op
     result = await revert_pay_for_request(test_key, integration_session, 500)
@@ -237,20 +246,28 @@ async def test_double_revert_prevented(
     This simulates the double-revert scenario where both upstream/base.py
     and proxy.py attempt to revert the same reservation.
     """
-    from routstr.auth import revert_pay_for_request
+    from routstr.auth import (
+        get_reservation_snapshot,
+        pay_for_request,
+        revert_pay_for_request,
+    )
 
     unique_key = f"test_double_revert_{uuid.uuid4().hex[:8]}"
     test_key = ApiKey(
         hashed_key=unique_key,
         balance=10000,
-        reserved_balance=500,
-        total_requests=5,
+        reserved_balance=0,
+        total_requests=4,
     )
     integration_session.add(test_key)
     await integration_session.commit()
+    await pay_for_request(test_key, 500, integration_session)
+    snapshot = await get_reservation_snapshot(test_key, integration_session)
 
     # First revert — should succeed
-    result1 = await revert_pay_for_request(test_key, integration_session, 500)
+    result1 = await revert_pay_for_request(
+        test_key, integration_session, 500, snapshot
+    )
     await integration_session.refresh(test_key)
 
     assert result1 is True
@@ -258,7 +275,9 @@ async def test_double_revert_prevented(
     assert test_key.total_requests == 4
 
     # Second revert of the same amount — should be no-op
-    result2 = await revert_pay_for_request(test_key, integration_session, 500)
+    result2 = await revert_pay_for_request(
+        test_key, integration_session, 500, snapshot
+    )
     await integration_session.refresh(test_key)
 
     assert result2 is False, "Second revert should be a no-op"
@@ -279,22 +298,30 @@ async def test_sequential_reverts_never_go_negative(
     Simulates the double-revert scenario where multiple code paths
     attempt to revert the same reservation.
     """
-    from routstr.auth import revert_pay_for_request
+    from routstr.auth import (
+        get_reservation_snapshot,
+        pay_for_request,
+        revert_pay_for_request,
+    )
 
     unique_key = f"test_multi_revert_{uuid.uuid4().hex[:8]}"
     test_key = ApiKey(
         hashed_key=unique_key,
         balance=10000,
-        reserved_balance=500,
-        total_requests=5,
+        reserved_balance=0,
+        total_requests=4,
     )
     integration_session.add(test_key)
     await integration_session.commit()
+    await pay_for_request(test_key, 500, integration_session)
+    snapshot = await get_reservation_snapshot(test_key, integration_session)
 
     # Run 5 sequential reverts for the same 500 reservation
     results = []
     for _ in range(5):
-        r = await revert_pay_for_request(test_key, integration_session, 500)
+        r = await revert_pay_for_request(
+            test_key, integration_session, 500, snapshot
+        )
         results.append(r)
 
     await integration_session.refresh(test_key)
@@ -317,7 +344,11 @@ async def test_child_key_revert_floor_guard(
     integration_session: AsyncSession,
 ) -> None:
     """Test that child key reserved_balance also has floor guard on revert."""
-    from routstr.auth import revert_pay_for_request
+    from routstr.auth import (
+        get_reservation_snapshot,
+        pay_for_request,
+        revert_pay_for_request,
+    )
 
     parent_key_hash = f"test_parent_{uuid.uuid4().hex[:8]}"
     child_key_hash = f"test_child_{uuid.uuid4().hex[:8]}"
@@ -325,22 +356,26 @@ async def test_child_key_revert_floor_guard(
     parent_key = ApiKey(
         hashed_key=parent_key_hash,
         balance=10000,
-        reserved_balance=500,
-        total_requests=3,
+        reserved_balance=0,
+        total_requests=2,
     )
     child_key = ApiKey(
         hashed_key=child_key_hash,
         balance=0,
-        reserved_balance=500,
-        total_requests=3,
+        reserved_balance=0,
+        total_requests=2,
         parent_key_hash=parent_key_hash,
     )
     integration_session.add(parent_key)
     integration_session.add(child_key)
     await integration_session.commit()
+    await pay_for_request(child_key, 500, integration_session)
+    snapshot = await get_reservation_snapshot(child_key, integration_session)
 
     # First revert succeeds
-    result1 = await revert_pay_for_request(child_key, integration_session, 500)
+    result1 = await revert_pay_for_request(
+        child_key, integration_session, 500, snapshot
+    )
     await integration_session.refresh(parent_key)
     await integration_session.refresh(child_key)
 
@@ -349,7 +384,9 @@ async def test_child_key_revert_floor_guard(
     assert child_key.reserved_balance == 0
 
     # Second revert is a no-op for both parent and child
-    result2 = await revert_pay_for_request(child_key, integration_session, 500)
+    result2 = await revert_pay_for_request(
+        child_key, integration_session, 500, snapshot
+    )
     await integration_session.refresh(parent_key)
     await integration_session.refresh(child_key)
 

tests/integration/test_reset_admin_password.py

@@ -0,0 +1,76 @@
+"""Tests for the ``reset_admin_password`` recovery script (issue #553).
+
+The script is the lockout escape hatch: it works without ``ROUTSTR_SECRET_KEY``
+(scrypt hashing is key-independent). Two explicit, mutually exclusive actions —
+``--password`` sets a new hash now, ``--regenerate`` clears the hash so the next
+boot generates and logs a fresh one. A bare invocation is informational only and
+must never touch the database (so nobody resets their password by accident).
+"""
+
+import pytest
+from sqlmodel.ext.asyncio.session import AsyncSession
+
+from routstr.core import vault
+from routstr.core.db import get_secret, set_admin_password
+from scripts.reset_admin_password import apply_reset, build_parser, main
+
+
+@pytest.mark.asyncio
+async def test_password_sets_a_verifiable_hash(
+    integration_session: AsyncSession,
+) -> None:
+    await apply_reset(integration_session, password="recover-me-123")
+
+    secret = await get_secret(integration_session)
+    assert secret.admin_password_hash is not None
+    assert vault.verify_password("recover-me-123", secret.admin_password_hash) is True
+    assert secret.updated_at is not None
+
+
+@pytest.mark.asyncio
+async def test_regenerate_clears_the_hash(
+    integration_session: AsyncSession,
+) -> None:
+    # Start from a node that already has an admin password set.
+    await set_admin_password(integration_session, "old-password-9")
+    assert (await get_secret(integration_session)).admin_password_hash is not None
+
+    await apply_reset(integration_session, regenerate=True)
+
+    secret = await get_secret(integration_session)
+    # Cleared -> the next boot's bootstrap_secrets generates and logs a new one.
+    assert secret.admin_password_hash is None
+    assert secret.updated_at is not None
+
+
+@pytest.mark.asyncio
+async def test_password_below_min_length_is_rejected(
+    integration_session: AsyncSession,
+) -> None:
+    await set_admin_password(integration_session, "old-password-9")
+
+    with pytest.raises(ValueError, match="8 characters"):
+        await apply_reset(integration_session, password="short")
+
+    # The existing password is untouched by the rejected reset.
+    secret = await get_secret(integration_session)
+    assert vault.verify_password("old-password-9", secret.admin_password_hash or "")
+
+
+def test_password_and_regenerate_are_mutually_exclusive() -> None:
+    parser = build_parser()
+    with pytest.raises(SystemExit):
+        parser.parse_args(["--password", "abcd1234", "--regenerate"])
+
+
+def test_no_args_prints_help_and_never_opens_a_session(
+    capsys: pytest.CaptureFixture[str],
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    def _fail() -> None:
+        raise AssertionError("a bare invocation must not touch the database")
+
+    monkeypatch.setattr("scripts.reset_admin_password.create_session", _fail)
+
+    assert main([]) == 0
+    assert "usage" in capsys.readouterr().out.lower()

tests/integration/test_secret_bootstrap.py

@@ -0,0 +1,484 @@
+"""Tests for ``bootstrap_secrets`` — moving node secrets into the Secret store.
+
+Specifies the per-secret bootstrap that runs at startup (issue #553). For both
+the admin password and the nsec it follows the same three branches: use the
+column if already set, otherwise migrate any legacy plaintext (env first, then
+the old settings blob), otherwise — admin password only — generate and log one.
+A column written under a different ROUTSTR_SECRET_KEY fails fast rather than
+silently corrupting state.
+"""
+
+import json
+from contextlib import asynccontextmanager
+from pathlib import Path
+from typing import Any, AsyncGenerator
+
+import pytest
+from sqlmodel import text
+from sqlmodel.ext.asyncio.session import AsyncSession
+
+from routstr.core import vault
+from routstr.core.db import NsecState, get_secret, set_nsec
+from routstr.core.settings import (
+    SettingsService,
+    bootstrap_secrets,
+    derive_npub_from_nsec,
+    settings,
+)
+
+# Valid Fernet keys; must match the suite default in tests/conftest.py.
+TEST_SECRET_KEY = "l_Tkp-7xmjcQ-IFhr6qhILrU8HPRbEmYMrfSbo_5srU="
+TEST_SECRET_KEY_ALT = "_Teyrky_iToeDK51Tj1FsI9MJ340_cqKGmeher-a7MQ="
+
+NSEC_HEX = "1" * 64
+# A different key, standing in for a stale value left behind in env/blob after
+# the vault has taken ownership of the real one.
+STALE_NSEC_HEX = "2" * 64
+
+
+@pytest.fixture
+def clean_secret_env(monkeypatch: pytest.MonkeyPatch) -> None:
+    """No ambient legacy secrets, and a known in-memory settings baseline."""
+    monkeypatch.delenv("ADMIN_PASSWORD", raising=False)
+    monkeypatch.delenv("NSEC", raising=False)
+    monkeypatch.setenv("ROUTSTR_SECRET_KEY", TEST_SECRET_KEY)
+    monkeypatch.setattr(settings, "nsec", "")
+    monkeypatch.setattr(settings, "npub", "")
+    monkeypatch.setattr(settings, "http_url", "")
+
+
+async def _create_settings_blob(session: AsyncSession, data: dict) -> None:
+    await session.exec(  # type: ignore
+        text(
+            "CREATE TABLE IF NOT EXISTS settings "
+            "(id INTEGER PRIMARY KEY, data TEXT NOT NULL, "
+            "updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP)"
+        )
+    )
+    await session.exec(  # type: ignore
+        text("INSERT INTO settings (id, data) VALUES (1, :data)").bindparams(
+            data=json.dumps(data)
+        )
+    )
+    await session.commit()
+
+
+# --- admin password --------------------------------------------------------
+
+
+@pytest.mark.asyncio
+async def test_generates_admin_password_when_none(
+    clean_secret_env: None, integration_session: AsyncSession
+) -> None:
+    await bootstrap_secrets(integration_session)
+    secret = await get_secret(integration_session)
+    assert secret.admin_password_hash is not None
+    assert secret.admin_password_hash.startswith("scrypt:")
+
+
+@pytest.mark.asyncio
+async def test_admin_password_generation_is_idempotent(
+    clean_secret_env: None, integration_session: AsyncSession
+) -> None:
+    await bootstrap_secrets(integration_session)
+    first = (await get_secret(integration_session)).admin_password_hash
+    await bootstrap_secrets(integration_session)
+    second = (await get_secret(integration_session)).admin_password_hash
+    assert first is not None and first == second
+
+
+@pytest.mark.asyncio
+async def test_hashes_legacy_admin_password_from_env(
+    clean_secret_env: None,
+    integration_session: AsyncSession,
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    monkeypatch.setenv("ADMIN_PASSWORD", "hunter2")
+    await bootstrap_secrets(integration_session)
+    secret = await get_secret(integration_session)
+    assert secret.admin_password_hash is not None
+    assert vault.verify_password("hunter2", secret.admin_password_hash) is True
+
+
+@pytest.mark.asyncio
+async def test_hashes_legacy_admin_password_from_blob(
+    clean_secret_env: None, integration_session: AsyncSession
+) -> None:
+    # No ADMIN_PASSWORD in env, but the old settings blob carries one.
+    await _create_settings_blob(integration_session, {"admin_password": "blobpw"})
+    await bootstrap_secrets(integration_session)
+    secret = await get_secret(integration_session)
+    assert vault.verify_password("blobpw", secret.admin_password_hash or "") is True
+
+
+@pytest.mark.asyncio
+async def test_admin_password_race_adopts_winner_without_clobber(
+    clean_secret_env: None,
+    integration_engine: Any,
+    integration_session: AsyncSession,
+    monkeypatch: pytest.MonkeyPatch,
+    capsys: pytest.CaptureFixture[str],
+) -> None:
+    # Two workers boot against one shared DB and both read a null admin password.
+    # The first to commit "wins" and shows the operator its generated password. A
+    # worker that read null but lost the race must NOT overwrite the winner's hash
+    # (which the operator may already be logging in with) and must NOT print a
+    # second password that will never work.
+    #
+    # The race window is forced deterministically: a hook fires inside bootstrap's
+    # generate branch (so it only runs once this worker has committed to
+    # generating) and commits the winner's password on a separate connection
+    # before this worker writes its own.
+    import sqlite3
+
+    from routstr.core import settings as settings_mod
+
+    db_file = integration_engine.url.database
+    winner_hash = vault.hash_password("winner-password-123")
+    real_token = settings_mod.secrets.token_urlsafe
+
+    def commit_winner_then_generate(nbytes: int) -> str:
+        conn = sqlite3.connect(db_file)
+        conn.execute(
+            "UPDATE secrets SET admin_password_hash = ? WHERE id = 1", (winner_hash,)
+        )
+        conn.commit()
+        conn.close()
+        return real_token(nbytes)
+
+    monkeypatch.setattr(
+        settings_mod.secrets, "token_urlsafe", commit_winner_then_generate
+    )
+
+    await get_secret(integration_session)  # row exists, password still null
+    capsys.readouterr()  # drop anything emitted before the race resolves
+    await bootstrap_secrets(integration_session)
+
+    secret = await get_secret(integration_session)
+    assert secret.admin_password_hash is not None
+    # The winner's password survives and still verifies — no clobber.
+    assert vault.verify_password("winner-password-123", secret.admin_password_hash)
+    # The losing worker stayed silent — no second generated password was leaked.
+    assert "generated a temporary" not in capsys.readouterr().out
+
+
+# --- nsec ------------------------------------------------------------------
+
+
+@pytest.mark.asyncio
+async def test_encrypts_legacy_nsec_from_env_and_derives_npub(
+    clean_secret_env: None,
+    integration_session: AsyncSession,
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    monkeypatch.setenv("NSEC", NSEC_HEX)
+    await bootstrap_secrets(integration_session)
+    secret = await get_secret(integration_session)
+    assert secret.encrypted_nsec is not None
+    assert vault.is_encrypted(secret.encrypted_nsec) is True
+    assert vault.decrypt(secret.encrypted_nsec) == NSEC_HEX
+    # In-memory runtime value is the decrypted nsec, and npub is derived from it.
+    assert settings.nsec == NSEC_HEX
+    assert settings.npub == derive_npub_from_nsec(NSEC_HEX)
+
+
+@pytest.mark.asyncio
+async def test_decrypts_existing_nsec_column(
+    clean_secret_env: None, integration_session: AsyncSession
+) -> None:
+    secret = await get_secret(integration_session)
+    secret.encrypted_nsec = vault.encrypt(NSEC_HEX)
+    secret.nsec_state = NsecState.encrypted
+    integration_session.add(secret)
+    await integration_session.commit()
+    stored = secret.encrypted_nsec
+
+    await bootstrap_secrets(integration_session)
+    reloaded = await get_secret(integration_session)
+    assert settings.nsec == NSEC_HEX
+    # The column is reused, not re-encrypted.
+    assert reloaded.encrypted_nsec == stored
+
+
+@pytest.mark.asyncio
+async def test_fail_fast_when_nsec_encrypted_with_different_key(
+    clean_secret_env: None,
+    integration_session: AsyncSession,
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    # Encrypt the column under the alternate key, then bootstrap under the
+    # suite key -> the value cannot be decrypted -> clear startup failure.
+    monkeypatch.setenv("ROUTSTR_SECRET_KEY", TEST_SECRET_KEY_ALT)
+    secret = await get_secret(integration_session)
+    secret.encrypted_nsec = vault.encrypt(NSEC_HEX)
+    secret.nsec_state = NsecState.encrypted
+    integration_session.add(secret)
+    await integration_session.commit()
+
+    monkeypatch.setenv("ROUTSTR_SECRET_KEY", TEST_SECRET_KEY)
+    with pytest.raises(RuntimeError, match="ROUTSTR_SECRET_KEY"):
+        await bootstrap_secrets(integration_session)
+
+
+# --- encryption is mandatory, key custody is not: upgrade without a key --------
+
+
+@pytest.mark.asyncio
+async def test_legacy_nsec_without_secret_key_generates_and_encrypts(
+    clean_secret_env: None,
+    integration_session: AsyncSession,
+    monkeypatch: pytest.MonkeyPatch,
+    tmp_path: Path,
+    capsys: pytest.CaptureFixture[str],
+) -> None:
+    # A node upgrading with a legacy plaintext NSEC but no ROUTSTR_SECRET_KEY must
+    # NOT break. Encryption at rest stays mandatory (the nsec is never persisted
+    # in plaintext), but the key custody is flexible: bootstrap generates a master
+    # key, persists it to the key file, warns loudly, and encrypts the identity —
+    # so the node keeps running instead of refusing to boot.
+    monkeypatch.delenv("ROUTSTR_SECRET_KEY", raising=False)
+    key_file = tmp_path / "routstr_secret.key"
+    monkeypatch.setenv("ROUTSTR_SECRET_KEY_FILE", str(key_file))
+    monkeypatch.setenv("NSEC", NSEC_HEX)
+
+    await bootstrap_secrets(integration_session)
+
+    # A master key was generated and persisted...
+    assert key_file.exists()
+    # ...the nsec is encrypted at rest under it, never stored in plaintext...
+    secret = await get_secret(integration_session)
+    assert secret.encrypted_nsec is not None
+    assert vault.is_encrypted(secret.encrypted_nsec) is True
+    assert vault.decrypt(secret.encrypted_nsec) == NSEC_HEX
+    assert secret.nsec_state == NsecState.encrypted
+    # ...the node holds the live identity (npub derived from it)...
+    assert settings.nsec == NSEC_HEX
+    assert settings.npub == derive_npub_from_nsec(NSEC_HEX)
+    # ...and the operator is loudly told a key was generated and must be backed up
+    # (path shown, but never the key value) so an upgrade cannot silently create
+    # an unbacked key nor leak the key into captured stdout / aggregated logs.
+    out = capsys.readouterr().out
+    assert str(key_file) in out
+    assert key_file.read_text().strip() not in out
+    assert "BACK UP" in out.upper()
+
+
+# --- boot ordering: rescue legacy blob secrets before they are stripped ----
+
+
+@pytest.mark.asyncio
+async def test_blob_only_nsec_is_migrated_before_blob_is_stripped(
+    clean_secret_env: None, integration_session: AsyncSession
+) -> None:
+    # Legacy node whose nsec lives ONLY in the settings blob (never in env).
+    # bootstrap_secrets must run *before* SettingsService.initialize strips the
+    # blob, or the only copy of the secret would be lost.
+    await _create_settings_blob(
+        integration_session, {"nsec": NSEC_HEX, "name": "LegacyNode"}
+    )
+
+    await bootstrap_secrets(integration_session)
+    await SettingsService.initialize(integration_session)
+
+    secret = await get_secret(integration_session)
+    # The plaintext nsec has been moved into the encrypted Secret store...
+    assert secret.encrypted_nsec is not None
+    assert vault.decrypt(secret.encrypted_nsec) == NSEC_HEX
+    assert settings.nsec == NSEC_HEX
+    # ...and stripped from the persisted settings blob.
+    row = await integration_session.exec(  # type: ignore
+        text("SELECT data FROM settings WHERE id = 1")
+    )
+    blob = json.loads(row.first()[0])
+    assert "nsec" not in blob
+    assert blob["name"] == "LegacyNode"
+
+
+@pytest.mark.asyncio
+async def test_initialize_does_not_clobber_store_only_nsec(
+    clean_secret_env: None, integration_session: AsyncSession
+) -> None:
+    # Steady state after migration: the nsec lives ONLY in the encrypted Secret
+    # store (NSEC removed from env, blob already stripped on a previous boot).
+    # bootstrap decrypts it into memory; initialize then re-derives settings from
+    # the secret-free blob and must NOT wipe the live nsec back to empty (or the
+    # node would silently stop signing Nostr announcements).
+    await _create_settings_blob(integration_session, {"name": "LegacyNode"})
+    secret = await get_secret(integration_session)
+    secret.encrypted_nsec = vault.encrypt(NSEC_HEX)
+    secret.nsec_state = NsecState.encrypted
+    integration_session.add(secret)
+    await integration_session.commit()
+
+    await bootstrap_secrets(integration_session)
+    assert settings.nsec == NSEC_HEX  # bootstrap decrypted it into memory
+
+    await SettingsService.initialize(integration_session)
+    # The live secret survives initialize even though no env/blob carries it...
+    assert settings.nsec == NSEC_HEX
+    # ...and is still never written back to the persisted blob.
+    row = await integration_session.exec(  # type: ignore
+        text("SELECT data FROM settings WHERE id = 1")
+    )
+    assert "nsec" not in json.loads(row.first()[0])
+
+
+@pytest.mark.asyncio
+async def test_stale_env_nsec_does_not_override_vault_nsec(
+    clean_secret_env: None,
+    integration_session: AsyncSession,
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    # The vault owns the nsec, but a stale NSEC (e.g. the operator rotated the
+    # key in the UI yet left the old value in .env) is still in the environment.
+    # bootstrap decrypts the store value; initialize must NOT let the stale env
+    # value clobber it, or a restart silently reverts to the old identity.
+    await set_nsec(integration_session, NSEC_HEX)
+
+    monkeypatch.setenv("NSEC", STALE_NSEC_HEX)
+    await _create_settings_blob(integration_session, {"name": "LegacyNode"})
+
+    await bootstrap_secrets(integration_session)
+    await SettingsService.initialize(integration_session)
+
+    # The vault value wins; the stale env value is ignored.
+    assert settings.nsec == NSEC_HEX
+
+
+@pytest.mark.asyncio
+async def test_stale_env_nsec_does_not_split_npub_from_vault_nsec(
+    clean_secret_env: None,
+    integration_session: AsyncSession,
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    # As above, the vault owns the nsec while a stale NSEC lingers in env. The
+    # private key correctly comes from the vault, but the npub must too: if
+    # initialize derives the public key from the stale env nsec, the node ends up
+    # with a private key from the vault and a public key from the old env value,
+    # and anything reading settings.npub announces the wrong Nostr identity.
+    expected_npub = derive_npub_from_nsec(NSEC_HEX)
+    stale_npub = derive_npub_from_nsec(STALE_NSEC_HEX)
+    assert expected_npub and stale_npub and expected_npub != stale_npub  # guard
+
+    await set_nsec(integration_session, NSEC_HEX)
+
+    monkeypatch.setenv("NSEC", STALE_NSEC_HEX)
+    await _create_settings_blob(integration_session, {"name": "LegacyNode"})
+
+    await bootstrap_secrets(integration_session)
+    await SettingsService.initialize(integration_session)
+
+    assert settings.nsec == NSEC_HEX
+    assert settings.npub == expected_npub
+
+
+@pytest.mark.asyncio
+async def test_cleared_nsec_stays_cleared_across_reboot(
+    clean_secret_env: None,
+    integration_session: AsyncSession,
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    # An identity was imported from env, then the operator cleared it via the
+    # admin API. The old NSEC is still in env. On the NEXT PROCESS the cleared
+    # identity must stay cleared, not get resurrected from the stale env value.
+    monkeypatch.setenv("NSEC", NSEC_HEX)
+    await bootstrap_secrets(integration_session)
+    assert settings.nsec == NSEC_HEX
+
+    # Clear via the admin path (store empty, vault owns it).
+    await set_nsec(integration_session, "")
+
+    # Simulate a fresh process rather than pre-clearing the live singleton: the
+    # pydantic settings global reloads the (still-stale) NSEC from env and derives
+    # its npub, which is exactly the in-memory state a new boot starts from before
+    # bootstrap runs. The cleared store must win over this stale live value.
+    monkeypatch.setattr(settings, "nsec", NSEC_HEX)
+    monkeypatch.setattr(settings, "npub", derive_npub_from_nsec(NSEC_HEX))
+
+    await bootstrap_secrets(integration_session)
+
+    reloaded = await get_secret(integration_session)
+    assert reloaded.nsec_state == NsecState.cleared
+    assert reloaded.encrypted_nsec is None  # not re-imported
+    assert settings.nsec == ""  # stays cleared
+    assert settings.npub == ""  # and no derived public identity survives
+
+
+@pytest.mark.asyncio
+async def test_initialize_keeps_npub_matching_store_only_nsec(
+    clean_secret_env: None, integration_session: AsyncSession
+) -> None:
+    # Steady state with mandatory encryption: the nsec lives ONLY in the
+    # encrypted Secret store (env carries no NSEC) and the blob has no npub.
+    # bootstrap decrypts the nsec and derives the npub into memory; initialize
+    # then re-derives settings from the npub-less blob and must NOT wipe the npub
+    # back to empty, or the node holds a private key with no matching public key
+    # and silently stops announcing a usable Nostr identity.
+    expected_npub = derive_npub_from_nsec(NSEC_HEX)
+    assert expected_npub  # guard: the test key must yield a real npub
+
+    await _create_settings_blob(integration_session, {"name": "LegacyNode"})
+    secret = await get_secret(integration_session)
+    secret.encrypted_nsec = vault.encrypt(NSEC_HEX)
+    secret.nsec_state = NsecState.encrypted
+    integration_session.add(secret)
+    await integration_session.commit()
+
+    await bootstrap_secrets(integration_session)
+    assert settings.npub == expected_npub  # bootstrap derived it
+
+    await SettingsService.initialize(integration_session)
+    # The npub still matches the live nsec...
+    assert settings.nsec == NSEC_HEX
+    assert settings.npub == expected_npub
+    # ...and is persisted to the blob (it is public, not a stripped secret).
+    row = await integration_session.exec(  # type: ignore
+        text("SELECT data FROM settings WHERE id = 1")
+    )
+    assert json.loads(row.first()[0])["npub"] == expected_npub
+
+
+@pytest.mark.asyncio
+async def test_startup_runs_bootstrap_before_settings_initialize(
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    # The two tests above prove the migration outcome *given* the call order;
+    # they hardcode that order themselves. This one guards the order at its real
+    # call site — the application lifespan — so a reorder in main.py (which would
+    # strip a blob-only secret before bootstrap could rescue it) is caught.
+    import routstr.core.main as main
+
+    order: list[str] = []
+
+    class _Abort(Exception):
+        pass
+
+    @asynccontextmanager
+    async def fake_create_session() -> AsyncGenerator[None, None]:
+        yield None
+
+    async def fake_bootstrap(session: Any) -> None:
+        order.append("bootstrap")
+
+    async def fake_initialize(session: Any) -> None:
+        order.append("initialize")
+        # Stop startup here, before the background-task fan-out (prices, nostr,
+        # upstreams) that we don't want to run in a unit test.
+        raise _Abort()
+
+    async def noop_init_db() -> None:
+        return None
+
+    monkeypatch.setattr(main, "configure_litellm", lambda: None)
+    monkeypatch.setattr(main, "register_deepseek_v4_pricing", lambda: None)
+    monkeypatch.setattr(main, "run_migrations", lambda: None)
+    monkeypatch.setattr(main, "init_db", noop_init_db)
+    monkeypatch.setattr(main, "create_session", fake_create_session)
+    monkeypatch.setattr(main, "bootstrap_secrets", fake_bootstrap)
+    monkeypatch.setattr(main.SettingsService, "initialize", fake_initialize)
+
+    with pytest.raises(_Abort):
+        async with main.lifespan(main.app):
+            pass
+
+    assert order == ["bootstrap", "initialize"]

tests/integration/test_secret_model.py

@@ -0,0 +1,93 @@
+"""Tests for the ``Secret`` singleton model (issue #553).
+
+Specifies the node-level secret store: a single row (``id=1``, like
+``RoutstrFee``) holding the one-way admin-password hash and the encrypted nsec.
+``get_secret`` is get-or-create, so callers always get the singleton without
+worrying whether it has been initialised yet. Encoding of the values themselves
+lives in ``routstr.core.vault``; here we only assert the row persists and stays
+a singleton.
+"""
+
+import time
+from typing import Any
+
+import pytest
+from sqlmodel import select
+from sqlmodel.ext.asyncio.session import AsyncSession
+
+from routstr.core.db import Secret, get_secret
+
+
+@pytest.mark.asyncio
+async def test_get_secret_creates_singleton(
+    integration_session: AsyncSession,
+) -> None:
+    secret = await get_secret(integration_session)
+    assert secret.id == 1
+    # Fresh row carries no secret material yet.
+    assert secret.admin_password_hash is None
+    assert secret.encrypted_nsec is None
+    assert secret.updated_at is None
+
+
+@pytest.mark.asyncio
+async def test_get_secret_is_idempotent(
+    integration_session: AsyncSession,
+) -> None:
+    first = await get_secret(integration_session)
+    second = await get_secret(integration_session)
+    assert first.id == second.id == 1
+    rows = (await integration_session.exec(select(Secret))).all()
+    assert len(rows) == 1
+
+
+@pytest.mark.asyncio
+async def test_secret_fields_round_trip(
+    integration_session: AsyncSession,
+) -> None:
+    secret = await get_secret(integration_session)
+    secret.admin_password_hash = "scrypt:16384:8:1:c2FsdA==:aGFzaA=="
+    secret.encrypted_nsec = "fernet:v1:gAAAAA"
+    secret.updated_at = int(time.time())
+    integration_session.add(secret)
+    await integration_session.commit()
+
+    integration_session.expunge_all()
+    reloaded = await get_secret(integration_session)
+    assert reloaded.admin_password_hash == "scrypt:16384:8:1:c2FsdA==:aGFzaA=="
+    assert reloaded.encrypted_nsec == "fernet:v1:gAAAAA"
+    assert reloaded.updated_at is not None
+
+
+@pytest.mark.asyncio
+async def test_get_secret_tolerates_concurrent_first_insert(
+    integration_engine: Any,
+    integration_session: AsyncSession,
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    # A second worker wins the race and commits the singleton row first.
+    async with AsyncSession(integration_engine, expire_on_commit=False) as other:
+        other.add(Secret(id=1, admin_password_hash="scrypt:from-other-worker"))
+        await other.commit()
+
+    # Reproduce the race window: our session's first read still sees no row, so
+    # it attempts to INSERT a duplicate id=1. The real IntegrityError that follows
+    # must be recovered (roll back, re-read) rather than crashing startup.
+    real_get = integration_session.get
+    calls = {"n": 0}
+
+    async def stale_first_read(model: Any, pk: Any) -> Any:
+        calls["n"] += 1
+        if calls["n"] == 1:
+            return None
+        return await real_get(model, pk)
+
+    monkeypatch.setattr(integration_session, "get", stale_first_read)
+
+    secret = await get_secret(integration_session)
+
+    # Recovered the other worker's row; no crash, still a single row.
+    assert secret.id == 1
+    assert secret.admin_password_hash == "scrypt:from-other-worker"
+    rows = (await integration_session.exec(select(Secret))).all()
+    assert len(rows) == 1

tests/integration/test_swap_fee_retry.py

@@ -0,0 +1,194 @@
+"""
+Integration tests for reactive swap fee retries via the wallet topup endpoint.
+
+Foreign-mint tokens are swapped to the primary mint using the foreign mint's
+melt quote, whose fee_reserve is a non-binding estimate (NUT-05): the mint may
+demand more when re-quoting or at melt execution. These tests cover the
+endpoint behaviour in those cases:
+
+1. The mint demands one sat more at melt time than every quote reported
+   (the mint.cubabitcoin.org incident): the swap retries with a smaller
+   invoice and the topup succeeds, crediting the recomputed amount.
+2. The real melt quote reports a higher fee_reserve than the estimate: the
+   swap re-quotes from the observed fee and the topup succeeds.
+3. The mint escalates its fee demands on every attempt: the retry budget is
+   exhausted and the endpoint returns 400 with a clear error (never 500),
+   without ever executing a melt.
+"""
+
+from collections.abc import Callable
+from unittest.mock import AsyncMock, Mock, patch
+
+import pytest
+from httpx import AsyncClient, Response
+
+from routstr.core.settings import settings
+
+# Captured at collection time, before the integration_app fixture replaces it
+# with the testmint stub that bypasses swapping (see conftest.py).
+from routstr.wallet import recieve_token as _real_recieve_token
+
+PRIMARY_MINT = "http://primary:3338"
+
+
+def _make_swap_mocks(
+    token_amount: int,
+    fee_reserves: list[int],
+    input_fees: int = 0,
+    mint_url: str = "http://foreign-mint:3338",
+) -> tuple[Mock, Mock, Mock]:
+    """Return (token, token_wallet, primary_wallet) mocks that act like a mint.
+
+    Mint quotes pass the requested amount through their ``request`` field and
+    melt quotes echo that amount back, so the mocks stay consistent for
+    whatever amounts the implementation requests. ``fee_reserves`` supplies the
+    fee_reserve of each successive melt quote (the first serves the estimation
+    pass); requesting more quotes than provided fails the test.
+    """
+    mock_token = Mock()
+    mock_token.mint = mint_url
+    mock_token.unit = "sat"
+    mock_token.amount = token_amount
+    mock_token.keysets = ["keyset1"]
+    mock_token.proofs = [Mock(amount=token_amount)]
+
+    mock_token_wallet = Mock()
+    mock_token_wallet.load_mint = AsyncMock()
+    mock_token_wallet.load_proofs = AsyncMock()
+    mock_token_wallet.get_fees_for_proofs = Mock(return_value=input_fees)
+
+    mock_primary_wallet = Mock()
+    mock_primary_wallet.load_mint = AsyncMock()
+    mock_primary_wallet.load_proofs = AsyncMock()
+    mock_primary_wallet.available_balance = Mock(amount=0)
+    mock_primary_wallet.mint = AsyncMock(return_value=Mock())
+
+    fees = iter(fee_reserves)
+
+    def _next_fee() -> int:
+        try:
+            return next(fees)
+        except StopIteration:
+            raise AssertionError(
+                "more melt quotes requested than fee_reserves provided"
+            ) from None
+
+    mock_primary_wallet.request_mint = AsyncMock(
+        side_effect=lambda amount: Mock(quote=f"mint_quote_{amount}", request=amount)
+    )
+    mock_token_wallet.melt_quote = AsyncMock(
+        side_effect=lambda invoice: Mock(
+            quote=f"melt_quote_{invoice}", amount=invoice, fee_reserve=_next_fee()
+        )
+    )
+    mock_token_wallet.melt = AsyncMock(return_value=Mock())
+
+    return mock_token, mock_token_wallet, mock_primary_wallet
+
+
+def _wallet_router(primary_wallet: Mock, token_wallet: Mock) -> Callable[..., Mock]:
+    """Route get_wallet calls to the primary or foreign wallet mock by URL."""
+
+    def fake_get_wallet(mint_url: str, unit: str = "sat", load: bool = True) -> Mock:
+        return primary_wallet if mint_url == PRIMARY_MINT else token_wallet
+
+    return fake_get_wallet
+
+
+async def _post_topup(
+    client: AsyncClient,
+    mock_token: Mock,
+    token_wallet: Mock,
+    primary_wallet: Mock,
+) -> Response:
+    """POST /v1/wallet/topup with the swap layer mocked at the mint boundary.
+
+    The conftest's testmint stub for recieve_token is swapped back for the
+    real implementation so the request exercises the actual swap path.
+    """
+    with patch("routstr.wallet.recieve_token", _real_recieve_token):
+        with patch(
+            "routstr.wallet.deserialize_token_from_string", return_value=mock_token
+        ):
+            with patch(
+                "routstr.wallet.get_wallet",
+                side_effect=_wallet_router(primary_wallet, token_wallet),
+            ):
+                with patch.object(settings, "primary_mint", PRIMARY_MINT):
+                    with patch.object(settings, "primary_mint_unit", "sat"):
+                        with patch.object(settings, "cashu_mints", [PRIMARY_MINT]):
+                            return await client.post(
+                                "/v1/wallet/topup",
+                                params={"cashu_token": "cashuAtest_foreign_token"},
+                            )
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_topup_retries_when_melt_demands_more_than_quoted(
+    authenticated_client: AsyncClient,
+) -> None:
+    """A 179-sat token where every quote reports fee_reserve=1 but the mint
+    rejects the first melt demanding 180. The retry shrinks the invoice to 177
+    and the topup credits 177 sats (177_000 msats)."""
+    mock_token, token_wallet, primary_wallet = _make_swap_mocks(
+        179, fee_reserves=[1, 1, 1], mint_url="http://mint.cubabitcoin.org"
+    )
+    token_wallet.melt.side_effect = [
+        Exception(
+            "Mint Error: not enough inputs provided for melt. "
+            "Provided: 179, needed: 180 (Code: 11000)"
+        ),
+        Mock(),
+    ]
+
+    response = await _post_topup(
+        authenticated_client, mock_token, token_wallet, primary_wallet
+    )
+
+    assert response.status_code == 200
+    assert response.json()["msats"] == 177_000
+    assert token_wallet.melt.call_count == 2
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_topup_retries_when_quote_fee_exceeds_estimate(
+    authenticated_client: AsyncClient,
+) -> None:
+    """A 1000-sat token estimated at fee 20, but the real quote demands 23.
+    The retry recomputes 1000 - 23 = 977, which fits, and the topup credits
+    977 sats (977_000 msats) with a single melt."""
+    mock_token, token_wallet, primary_wallet = _make_swap_mocks(
+        1000, fee_reserves=[20, 23, 23]
+    )
+
+    response = await _post_topup(
+        authenticated_client, mock_token, token_wallet, primary_wallet
+    )
+
+    assert response.status_code == 200
+    assert response.json()["msats"] == 977_000
+    assert token_wallet.melt.call_count == 1
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_topup_returns_422_when_retries_exhausted(
+    authenticated_client: AsyncClient,
+) -> None:
+    """A mint that escalates fee_reserve on every re-quote (1 → 10 → 25 → 50)
+    exhausts the retry budget: clean 422 mint_error/too-small taxonomy, melt
+    never executed."""
+    mock_token, token_wallet, primary_wallet = _make_swap_mocks(
+        1000, fee_reserves=[1, 10, 25, 50]
+    )
+
+    response = await _post_topup(
+        authenticated_client, mock_token, token_wallet, primary_wallet
+    )
+
+    assert response.status_code == 422
+    assert "too small to cover swap fees" in response.json()["detail"]
+    assert token_wallet.melt_quote.call_count == 4  # estimation + 3 attempts
+    token_wallet.melt.assert_not_called()

tests/integration/test_wallet_authentication.py

@@ -12,10 +12,7 @@ from sqlmodel import select
 
 from routstr.core.db import ApiKey
 
-from .utils import (
-    CashuTokenGenerator,
-    ResponseValidator,
-)
+from .utils import ResponseValidator
 
 
 @pytest.mark.integration
@@ -79,29 +76,31 @@ async def test_api_key_generation_invalid_token(
     # Capture initial state
     await db_snapshot.capture()
 
-    # Test various invalid tokens
-    invalid_tokens = [
-        CashuTokenGenerator.generate_invalid_token(),  # Malformed token
-        "not-a-cashu-token",  # Wrong format
-        "cashuA",  # Empty token
-        "cashuA" + "x" * 1000,  # Invalid base64
+    # Non-Cashu bearer values are invalid API keys (401). Malformed values that
+    # look like Cashu tokens use the shared Cashu taxonomy (400 invalid_token).
+    invalid_tokens: list[tuple[str, int, str | None]] = [
+        ("not-a-cashu-token", 401, None),
+        ("sk-not-a-real-api-key", 401, None),
+        ("cashuA", 400, "invalid_cashu_token"),
+        ("cashuA" + "x" * 1000, 400, "invalid_cashu_token"),
     ]
 
-    for invalid_token in invalid_tokens:
+    for invalid_token, expected_status, expected_code in invalid_tokens:
         integration_client.headers["Authorization"] = f"Bearer {invalid_token}"
         response = await integration_client.get("/v1/wallet/info")
 
-        # Should fail with 401
-        assert response.status_code == 401, (
+        assert response.status_code == expected_status, (
             f"Token {invalid_token[:20]}... should be invalid"
         )
 
         # Validate error response
         validator = ResponseValidator()
         error_validation = validator.validate_error_response(
-            response, expected_status=401, expected_error_key="detail"
+            response, expected_status=expected_status, expected_error_key="detail"
         )
         assert error_validation["valid"]
+        if expected_code is not None:
+            assert response.json()["detail"]["error"]["code"] == expected_code
 
     # Verify no database changes
     diff = await db_snapshot.diff()

tests/integration/test_wallet_refund.py

@@ -14,6 +14,7 @@ from httpx import AsyncClient
 from sqlmodel import select
 
 from routstr.core.db import ApiKey, CashuTransaction
+from routstr.wallet import MintConnectionError
 
 
 @pytest.mark.integration
@@ -503,26 +504,19 @@ async def test_mint_unavailability_handling(
 
     # The global mock in conftest.py is already in place,
     # so we need to temporarily modify it
-    from unittest.mock import patch
+    raw_error = "Mint unavailable: Connection refused"
 
-    # Make the send_token method raise an exception
+    # Make the send_token method raise a typed mint connection exception.
     with patch(
         "routstr.balance.send_token",
-        side_effect=Exception("Mint unavailable: Connection refused"),
+        side_effect=MintConnectionError(raw_error),
     ):
-        # The exception should propagate as a 503 error (Service Unavailable)
-        # But we need to handle it properly
-        try:
-            response = await authenticated_client.post("/v1/wallet/refund")
-            # If we get here, check the status code
-            assert response.status_code == 503
-            assert "Mint service unavailable" in response.json()["detail"]
-        except Exception as e:
-            # If the exception propagates, that's also a failure scenario
-            assert "Mint unavailable" in str(e)
+        response = await authenticated_client.post("/v1/wallet/refund")
+        assert response.status_code == 503
+        assert response.json()["detail"] == "Mint service unavailable"
+        assert raw_error not in response.text
 
         # Balance should remain unchanged (transaction should roll back)
-        # Note: Current implementation might not handle this perfectly
         wallet_response = await authenticated_client.get("/v1/wallet/")
         assert wallet_response.status_code == 200
         assert wallet_response.json()["balance"] == 10_000_000

tests/unit/test_admin_transactions.py

@@ -0,0 +1,35 @@
+from contextlib import asynccontextmanager
+from unittest.mock import AsyncMock, MagicMock, patch
+
+import pytest
+
+from routstr.core.admin import get_transactions_api
+from routstr.core.db import CashuTransaction
+
+
+@pytest.mark.asyncio
+async def test_transactions_api_excludes_internal_sweep_claim_timestamp() -> None:
+    transaction = CashuTransaction(
+        token="cashu-token",
+        amount=10,
+        unit="sat",
+        type="out",
+        sweep_started_at=123,
+    )
+    count_result = MagicMock()
+    count_result.one.return_value = 1
+    transactions_result = MagicMock()
+    transactions_result.all.return_value = [transaction]
+    session = MagicMock()
+    session.exec = AsyncMock(side_effect=[count_result, transactions_result])
+
+    @asynccontextmanager
+    async def create_session():  # type: ignore[no-untyped-def]
+        yield session
+
+    with patch("routstr.core.admin.create_session", create_session):
+        response = await get_transactions_api()
+
+    assert response["total"] == 1
+    assert response["transactions"][0]["token"] == "cashu-token"
+    assert "sweep_started_at" not in response["transactions"][0]

tests/unit/test_admin_withdraw.py

@@ -0,0 +1,82 @@
+from types import SimpleNamespace
+from unittest.mock import AsyncMock, Mock
+
+import pytest
+
+from routstr.core import admin
+
+
+@pytest.mark.asyncio
+@pytest.mark.parametrize("requested_mint", [None, "https://secondary.example"])
+async def test_withdraw_uses_effective_mint_and_records_outgoing_transaction(
+    monkeypatch: pytest.MonkeyPatch, requested_mint: str | None
+) -> None:
+    primary_mint = "https://primary.example"
+    effective_mint = requested_mint or primary_mint
+    wallet = object()
+    proofs = [SimpleNamespace(amount=40), SimpleNamespace(amount=60)]
+    token = "cashuBoutgoing"
+
+    get_wallet = AsyncMock(return_value=wallet)
+    get_proofs = Mock(return_value=proofs)
+    filter_proofs = AsyncMock(return_value=proofs)
+    send_token = AsyncMock(return_value=token)
+    store_transaction = AsyncMock(return_value=True)
+
+    monkeypatch.setattr(admin, "get_wallet", get_wallet)
+    monkeypatch.setattr(admin, "get_proofs_per_mint_and_unit", get_proofs)
+    monkeypatch.setattr(admin, "slow_filter_spend_proofs", filter_proofs)
+    monkeypatch.setattr(admin, "send_token", send_token)
+    monkeypatch.setattr(admin, "store_cashu_transaction", store_transaction)
+    monkeypatch.setattr(admin.settings, "primary_mint", primary_mint)
+
+    result = await admin.withdraw(
+        Mock(),
+        admin.WithdrawRequest(amount=75, mint_url=requested_mint, unit="sat"),
+    )
+
+    assert result == {"token": token}
+    get_wallet.assert_awaited_once_with(effective_mint, "sat")
+    get_proofs.assert_called_once_with(wallet, effective_mint, "sat", not_reserved=True)
+    filter_proofs.assert_awaited_once_with(proofs, wallet)
+    send_token.assert_awaited_once_with(75, "sat", effective_mint)
+    store_transaction.assert_awaited_once_with(
+        token=token,
+        amount=75,
+        unit="sat",
+        mint_url=effective_mint,
+        typ="out",
+        collected=False,
+        source="admin",
+    )
+
+
+@pytest.mark.asyncio
+async def test_withdraw_returns_issued_token_when_audit_storage_fails(
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    mint = "https://primary.example"
+    proofs = [SimpleNamespace(amount=100)]
+    token = "cashuBrecoverable"
+
+    monkeypatch.setattr(admin, "get_wallet", AsyncMock(return_value=object()))
+    monkeypatch.setattr(
+        admin, "get_proofs_per_mint_and_unit", Mock(return_value=proofs)
+    )
+    monkeypatch.setattr(
+        admin, "slow_filter_spend_proofs", AsyncMock(return_value=proofs)
+    )
+    monkeypatch.setattr(admin, "send_token", AsyncMock(return_value=token))
+    monkeypatch.setattr(
+        admin,
+        "store_cashu_transaction",
+        AsyncMock(side_effect=RuntimeError("database unavailable")),
+    )
+    critical = Mock()
+    monkeypatch.setattr(admin.logger, "critical", critical)
+    monkeypatch.setattr(admin.settings, "primary_mint", mint)
+
+    result = await admin.withdraw(Mock(), admin.WithdrawRequest(amount=75))
+
+    assert result == {"token": token}
+    critical.assert_called_once()

tests/unit/test_algorithm.py

@@ -137,12 +137,12 @@ def test_create_model_mappings_includes_db_override_for_missing_cached_model(
 
     model_instances, provider_map, unique_models = create_model_mappings(
         upstreams=[provider],
-        overrides_by_id={"azure/gpt-4o": (override_row, 1.01)},
-        disabled_model_ids=set(),
+        overrides_by_key={("azure/gpt-4o", 7): (override_row, 1.01)},
+        disabled_model_keys=set(),
     )
 
     assert "azure/gpt-4o" in model_instances
-    assert provider_map["azure/gpt-4o"] == [provider]
+    assert [p for _, p in provider_map["azure/gpt-4o"]] == [provider]
     assert "gpt-4o" in unique_models
 
 
@@ -182,11 +182,73 @@ def test_create_model_mappings_dedupes_with_provider_identity_not_provider_type(
 
     _, provider_map, _ = create_model_mappings(
         upstreams=[provider_a, provider_b],
-        overrides_by_id={"azure/gpt-4o": (override_row, 1.01)},
-        disabled_model_ids=set(),
+        overrides_by_key={("azure/gpt-4o", 2): (override_row, 1.01)},
+        disabled_model_keys=set(),
     )
 
-    providers_for_alias = provider_map["azure/gpt-4o"]
+    providers_for_alias = [p for _, p in provider_map["azure/gpt-4o"]]
     assert provider_a in providers_for_alias
     assert provider_b in providers_for_alias
     assert len(providers_for_alias) == 2
+
+
+def test_create_model_mappings_applies_override_only_to_matching_provider(
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    """Same-id overrides must not add provider-specific aliases to other providers."""
+    provider_a_model = create_test_model("same-id", prompt_price=0.01)
+    provider_a = create_test_provider(
+        "provider-a",
+        "https://provider-a.example/v1",
+        db_id=1,
+        models=[provider_a_model],
+    )
+    provider_b_model = create_test_model("same-id", prompt_price=0.02)
+    provider_b = create_test_provider(
+        "provider-b",
+        "https://provider-b.example/v1",
+        db_id=2,
+        models=[provider_b_model],
+    )
+
+    override_model = create_test_model("same-id", prompt_price=0.001)
+    override_model.alias_ids = ["provider-b-only"]
+    override_row = SimpleNamespace(id="same-id", upstream_provider_id=2, enabled=True)
+
+    def fake_row_to_model(*args, **kwargs) -> Model:  # type: ignore[no-untyped-def]
+        return override_model
+
+    monkeypatch.setattr("routstr.payment.models._row_to_model", fake_row_to_model)
+
+    _, provider_map, _ = create_model_mappings(
+        upstreams=[provider_a, provider_b],
+        overrides_by_key={("same-id", 2): (override_row, 1.01)},
+        disabled_model_keys=set(),
+    )
+
+    assert [p for _, p in provider_map["provider-b-only"]] == [provider_b]
+    assert {p for _, p in provider_map["same-id"]} == {provider_a, provider_b}
+
+
+def test_create_model_mappings_disables_only_matching_provider() -> None:
+    """Disabled overrides are scoped to the provider row, not the shared model id."""
+    provider_a = create_test_provider(
+        "provider-a",
+        "https://provider-a.example/v1",
+        db_id=1,
+        models=[create_test_model("same-id")],
+    )
+    provider_b = create_test_provider(
+        "provider-b",
+        "https://provider-b.example/v1",
+        db_id=2,
+        models=[create_test_model("same-id")],
+    )
+
+    _, provider_map, _ = create_model_mappings(
+        upstreams=[provider_a, provider_b],
+        overrides_by_key={},
+        disabled_model_keys={("same-id", 2)},
+    )
+
+    assert [p for _, p in provider_map["same-id"]] == [provider_a]

tests/unit/test_auth_cashu.py

@@ -0,0 +1,293 @@
+import hashlib
+from types import SimpleNamespace
+from typing import AsyncGenerator, cast
+from unittest.mock import AsyncMock, patch
+
+import httpx
+import pytest
+from fastapi import HTTPException
+from sqlalchemy.ext.asyncio import AsyncEngine, create_async_engine
+from sqlalchemy.pool import StaticPool
+from sqlmodel import SQLModel
+from sqlmodel.ext.asyncio.session import AsyncSession
+
+from routstr.auth import validate_bearer_key
+from routstr.core.db import ApiKey
+from routstr.wallet import MintConnectionError
+
+
+def _value_error_wrapping_transport() -> ValueError:
+    """A ValueError re-raised ``from`` a real httpx transport error, mirroring
+    ``wallet.py`` wrapping a connection failure. The sanitized classifier must
+    still see the mint-unreachable signal through the ``__cause__`` chain."""
+    try:
+        raise httpx.ConnectError("All connection attempts failed")
+    except httpx.ConnectError as exc:
+        err = ValueError("Failed to estimate fees: connection failed")
+        err.__cause__ = exc
+        return err
+
+
+def _make_engine() -> AsyncEngine:
+    return create_async_engine(
+        "sqlite+aiosqlite://",
+        poolclass=StaticPool,
+        connect_args={"check_same_thread": False},
+    )
+
+
+@pytest.fixture
+async def session() -> AsyncGenerator[AsyncSession, None]:
+    engine = _make_engine()
+    async with engine.begin() as conn:
+        await conn.run_sync(SQLModel.metadata.create_all)
+    db_session = AsyncSession(engine, expire_on_commit=False)
+    try:
+        yield db_session
+    finally:
+        await db_session.close()
+        await engine.dispose()
+
+
+@pytest.mark.asyncio
+async def test_failed_first_cashu_redemption_rolls_back_empty_api_key(
+    session: AsyncSession,
+) -> None:
+    token = "cashuAfirst_seen_but_redemption_fails"
+    hashed_key = hashlib.sha256(token.encode()).hexdigest()
+    token_obj = SimpleNamespace(mint="http://mint:3338", unit="sat")
+
+    from routstr.core.settings import settings
+
+    with (
+        patch.object(settings, "cashu_mints", ["http://mint:3338"]),
+        patch("routstr.auth.deserialize_token_from_string", return_value=token_obj),
+        patch(
+            "routstr.auth.credit_balance",
+            new=AsyncMock(side_effect=ValueError("token already spent")),
+        ),
+    ):
+        with pytest.raises(HTTPException):
+            await validate_bearer_key(token, session)
+
+    assert await session.get(ApiKey, hashed_key) is None
+
+
+@pytest.mark.parametrize(
+    ("error", "expected_status", "expected_type", "expected_message", "expected_code"),
+    [
+        (
+            ValueError("Mint Error: Token already spent. (Code: 11001)"),
+            400,
+            "token_already_spent",
+            "Cashu token already spent",
+            "cashu_token_already_spent",
+        ),
+        (
+            # Raw httpx transport error propagated unwrapped from cashu.
+            httpx.ConnectError("All connection attempts failed"),
+            503,
+            "mint_unreachable",
+            "Cashu mint is unreachable",
+            "cashu_mint_unreachable",
+        ),
+        (
+            # Typed error raised by wallet.py at a wrap site.
+            MintConnectionError("connect to http://mint:3338 refused"),
+            503,
+            "mint_unreachable",
+            "Cashu mint is unreachable",
+            "cashu_mint_unreachable",
+        ),
+        (
+            # ValueError wrapping the httpx error in its __cause__ chain.
+            _value_error_wrapping_transport(),
+            503,
+            "mint_unreachable",
+            "Cashu mint is unreachable",
+            "cashu_mint_unreachable",
+        ),
+        (
+            # asyncio.TimeoutError is builtin TimeoutError on 3.11+.
+            TimeoutError("Timed out connecting to Cashu mint http://mint:3338"),
+            503,
+            "mint_unreachable",
+            "Cashu mint is unreachable",
+            "cashu_mint_unreachable",
+        ),
+        (
+            ValueError(
+                "Token amount (5 sat) is insufficient to cover melt fees. "
+                "Needed: 7 sat (amount: 5 + fee: 1 + input_fees: 1)"
+            ),
+            422,
+            "mint_error",
+            "Token value is too small to cover swap fees",
+            "cashu_token_swap_fees_exceed_amount",
+        ),
+        (
+            ValueError(
+                "Failed to estimate fees: Fees (7 sat) exceed token amount (5 sat)"
+            ),
+            422,
+            "mint_error",
+            "Token value is too small to cover swap fees",
+            "cashu_token_swap_fees_exceed_amount",
+        ),
+        (
+            ValueError(
+                "Failed to melt token from foreign mint http://foreign:3338: boom"
+            ),
+            422,
+            "mint_error",
+            "Failed to swap token from foreign mint",
+            "cashu_foreign_mint_swap_failed",
+        ),
+        (
+            ValueError("could not decode token"),
+            400,
+            "invalid_token",
+            "Invalid Cashu token",
+            "invalid_cashu_token",
+        ),
+        (
+            ValueError("some unexpected wallet condition"),
+            400,
+            "cashu_error",
+            "Failed to redeem Cashu token",
+            "cashu_token_redemption_failed",
+        ),
+        (
+            ValueError("Redeemed token amount must be positive, got 0 msats"),
+            400,
+            "cashu_error",
+            "Failed to redeem Cashu token: token yielded no value",
+            "cashu_token_zero_value",
+        ),
+    ],
+)
+@pytest.mark.asyncio
+async def test_redemption_failure_returns_sanitized_error(
+    session: AsyncSession,
+    error: Exception,
+    expected_status: int,
+    expected_type: str,
+    expected_message: str,
+    expected_code: str,
+) -> None:
+    """Redemption failures reuse the shared X-Cashu taxonomy (carried in
+    ``type``), expose stable sanitized messages and granular machine-readable
+    ``code`` values, and leave no orphan ApiKey row."""
+    token = "cashuAredemption_fails_with_specific_error"
+    hashed_key = hashlib.sha256(token.encode()).hexdigest()
+    token_obj = SimpleNamespace(mint="http://mint:3338", unit="sat")
+
+    from routstr.core.settings import settings
+
+    with (
+        patch.object(settings, "cashu_mints", ["http://mint:3338"]),
+        patch("routstr.auth.deserialize_token_from_string", return_value=token_obj),
+        patch(
+            "routstr.auth.credit_balance",
+            new=AsyncMock(side_effect=error),
+        ),
+    ):
+        with pytest.raises(HTTPException) as exc_info:
+            await validate_bearer_key(token, session)
+
+    assert exc_info.value.status_code == expected_status
+    detail = cast(dict[str, dict[str, object]], exc_info.value.detail)
+    error_detail = detail["error"]
+    assert error_detail["type"] == expected_type
+    assert error_detail["code"] == expected_code
+    assert error_detail["message"] == expected_message
+    assert str(error) not in cast(str, error_detail["message"])
+    assert await session.get(ApiKey, hashed_key) is None
+
+
+@pytest.mark.asyncio
+async def test_unexpected_redemption_error_returns_internal_error(
+    session: AsyncSession,
+) -> None:
+    """Unexpected (non-wallet) failures surface as generic 500s without
+    leaking internal details, instead of masquerading as token errors."""
+    token = "cashuAredemption_fails_with_internal_error"
+    hashed_key = hashlib.sha256(token.encode()).hexdigest()
+    token_obj = SimpleNamespace(mint="http://mint:3338", unit="sat")
+
+    from routstr.core.settings import settings
+
+    with (
+        patch.object(settings, "cashu_mints", ["http://mint:3338"]),
+        patch("routstr.auth.deserialize_token_from_string", return_value=token_obj),
+        patch(
+            "routstr.auth.credit_balance",
+            new=AsyncMock(side_effect=RuntimeError("db exploded at /var/lib/secret")),
+        ),
+    ):
+        with pytest.raises(HTTPException) as exc_info:
+            await validate_bearer_key(token, session)
+
+    assert exc_info.value.status_code == 500
+    detail = cast(dict[str, dict[str, str]], exc_info.value.detail)
+    error_detail = detail["error"]
+    assert error_detail["code"] == "internal_error"
+    assert "/var/lib/secret" not in error_detail["message"]
+    assert await session.get(ApiKey, hashed_key) is None
+
+
+@pytest.mark.asyncio
+async def test_internal_error_with_invalid_keyword_does_not_masquerade(
+    session: AsyncSession,
+) -> None:
+    """A non-wallet fault whose text merely contains "invalid" (but not
+    "token") must fall through to a generic 500, not a 401 token error.
+
+    Guards the anchored `"invalid"/"decode"` + `"token"` gate against stdlib/
+    driver strings like "Invalid isoformat string" leaking as token errors."""
+    token = "cashuAinternal_fault_mentions_invalid"
+    hashed_key = hashlib.sha256(token.encode()).hexdigest()
+    token_obj = SimpleNamespace(mint="http://mint:3338", unit="sat")
+
+    from routstr.core.settings import settings
+
+    with (
+        patch.object(settings, "cashu_mints", ["http://mint:3338"]),
+        patch("routstr.auth.deserialize_token_from_string", return_value=token_obj),
+        patch(
+            "routstr.auth.credit_balance",
+            new=AsyncMock(
+                side_effect=RuntimeError("Invalid isoformat string: '2020-13-99'")
+            ),
+        ),
+    ):
+        with pytest.raises(HTTPException) as exc_info:
+            await validate_bearer_key(token, session)
+
+    assert exc_info.value.status_code == 500
+    detail = cast(dict[str, dict[str, str]], exc_info.value.detail)
+    assert detail["error"]["code"] == "internal_error"
+    assert await session.get(ApiKey, hashed_key) is None
+
+
+@pytest.mark.asyncio
+async def test_malformed_cashu_token_returns_400_invalid_token(
+    session: AsyncSession,
+) -> None:
+    """A malformed 'cashu...' token that fails to decode maps to 400
+    invalid_cashu_token (shared taxonomy), not the generic 401 invalid_api_key."""
+    token = "cashuAthis_is_not_a_valid_token"
+
+    with patch(
+        "routstr.auth.deserialize_token_from_string",
+        side_effect=ValueError("unable to decode token: bad base64"),
+    ):
+        with pytest.raises(HTTPException) as exc_info:
+            await validate_bearer_key(token, session)
+
+    assert exc_info.value.status_code == 400
+    detail = cast(dict[str, dict[str, str]], exc_info.value.detail)
+    assert detail["error"]["type"] == "invalid_token"
+    assert detail["error"]["code"] == "invalid_cashu_token"
+    # Raw decoder text must not leak to the client.
+    assert "base64" not in detail["error"]["message"]

tests/unit/test_auto_topup.py

@@ -0,0 +1,143 @@
+import json
+from unittest.mock import AsyncMock, MagicMock, patch
+
+import pytest
+
+from routstr.core.db import CashuTransaction
+from routstr.upstream.auto_topup import _check_and_topup
+
+
+def _row() -> MagicMock:
+    row = MagicMock()
+    row.id = "provider-1"
+    row.base_url = "https://provider.test"
+    row.api_key = "secret"
+    row.provider_settings = json.dumps(
+        {
+            "auto_topup": True,
+            "topup_threshold": 100,
+            "topup_amount_limit": 50,
+            "topup_mint_url": "https://mint.test",
+        }
+    )
+    return row
+
+
+class _Session:
+    def __init__(self, transaction: CashuTransaction) -> None:
+        self.transaction = transaction
+        self.commit = AsyncMock()
+
+    async def __aenter__(self) -> "_Session":
+        return self
+
+    async def __aexit__(self, *args: object) -> None:
+        return None
+
+    async def exec(self, query: object) -> MagicMock:
+        result = MagicMock()
+        result.first.return_value = self.transaction
+        return result
+
+    def add(self, transaction: CashuTransaction) -> None:
+        self.transaction = transaction
+
+
+@pytest.mark.asyncio
+async def test_auto_topup_persists_before_sending_and_marks_success_collected() -> None:
+    provider = MagicMock()
+    provider.get_balance = AsyncMock(return_value=0)
+    provider.topup = AsyncMock(return_value={"balance": 50})
+    transaction = CashuTransaction(
+        token="cashu-token", amount=50, unit="sat", source="auto_topup"
+    )
+    session = _Session(transaction)
+
+    with (
+        patch(
+            "routstr.upstream.auto_topup.RoutstrUpstreamProvider.from_db_row",
+            return_value=provider,
+        ),
+        patch(
+            "routstr.upstream.auto_topup.send_token",
+            AsyncMock(return_value="cashu-token"),
+        ),
+        patch(
+            "routstr.upstream.auto_topup.store_cashu_transaction",
+            AsyncMock(return_value=True),
+        ) as store,
+        patch("routstr.upstream.auto_topup.create_session", return_value=session),
+    ):
+        await _check_and_topup(_row())
+
+    store.assert_awaited_once_with(
+        token="cashu-token",
+        amount=50,
+        unit="sat",
+        mint_url="https://mint.test",
+        typ="out",
+        collected=False,
+        source="auto_topup",
+    )
+    provider.topup.assert_awaited_once_with("cashu-token")
+    assert transaction.collected is True
+    session.commit.assert_awaited_once()
+
+
+@pytest.mark.asyncio
+@pytest.mark.parametrize("outcome", [{"error": "rejected"}, RuntimeError("network")])
+async def test_auto_topup_failure_leaves_persisted_token_uncollected(
+    outcome: object,
+) -> None:
+    provider = MagicMock()
+    provider.get_balance = AsyncMock(return_value=0)
+    provider.topup = AsyncMock(
+        side_effect=outcome if isinstance(outcome, Exception) else None,
+        return_value=outcome,
+    )
+
+    with (
+        patch(
+            "routstr.upstream.auto_topup.RoutstrUpstreamProvider.from_db_row",
+            return_value=provider,
+        ),
+        patch(
+            "routstr.upstream.auto_topup.send_token",
+            AsyncMock(return_value="cashu-token"),
+        ),
+        patch(
+            "routstr.upstream.auto_topup.store_cashu_transaction",
+            AsyncMock(return_value=True),
+        ),
+        patch("routstr.upstream.auto_topup.create_session") as create_session,
+    ):
+        if isinstance(outcome, Exception):
+            with pytest.raises(RuntimeError):
+                await _check_and_topup(_row())
+        else:
+            await _check_and_topup(_row())
+
+    create_session.assert_not_called()
+
+
+@pytest.mark.asyncio
+async def test_auto_topup_does_not_send_untracked_token() -> None:
+    provider = MagicMock()
+    provider.get_balance = AsyncMock(return_value=0)
+    provider.topup = AsyncMock()
+    with (
+        patch(
+            "routstr.upstream.auto_topup.RoutstrUpstreamProvider.from_db_row",
+            return_value=provider,
+        ),
+        patch(
+            "routstr.upstream.auto_topup.send_token",
+            AsyncMock(return_value="cashu-token"),
+        ),
+        patch(
+            "routstr.upstream.auto_topup.store_cashu_transaction",
+            AsyncMock(side_effect=RuntimeError("database unavailable")),
+        ),
+    ):
+        await _check_and_topup(_row())
+    provider.topup.assert_not_awaited()

tests/unit/test_balance.py

@@ -1,12 +1,13 @@
 import json
 from unittest.mock import AsyncMock, MagicMock, patch
 
+import httpx
 import pytest
 from fastapi.responses import JSONResponse
 
-from routstr.balance import refund_wallet_endpoint
+from routstr.balance import refund_wallet_endpoint, topup_wallet_endpoint
 from routstr.core.db import ApiKey, CashuTransaction
-from routstr.wallet import credit_balance
+from routstr.wallet import MintConnectionError, credit_balance
 
 
 def _make_cashu_tx(
@@ -103,6 +104,64 @@ async def test_refund_x_cashu_not_found_raises_404() -> None:
     assert exc_info.value.status_code == 404
 
 
+@pytest.mark.asyncio
+async def test_refund_x_cashu_pending_raises_425() -> None:
+    """in row exists with a request_id but out row not yet created → 425.
+
+    This is the race condition where /v1/wallet/refund is polled while the
+    upstream request is still in flight. The endpoint must signal "retry"
+    rather than a permanent 404.
+    """
+    from fastapi import HTTPException
+
+    x_cashu_token = "cashuApending_token"
+    in_tx = _make_cashu_tx(
+        token=x_cashu_token, amount=0, unit="msat", type="in", request_id="req-pending"
+    )
+
+    session = MagicMock()
+    session.exec = AsyncMock(side_effect=[_exec_result(in_tx), _exec_result(None)])
+    session.add = MagicMock()
+    session.commit = AsyncMock()
+
+    with pytest.raises(HTTPException) as exc_info:
+        await refund_wallet_endpoint(
+            authorization="Bearer sk-somekey",
+            x_cashu=x_cashu_token,
+            session=session,
+        )
+
+    assert exc_info.value.status_code == 425
+    assert exc_info.value.headers == {"Retry-After": "2"}
+
+
+@pytest.mark.asyncio
+async def test_refund_x_cashu_in_tx_without_request_id_raises_404() -> None:
+    """in row exists but has no request_id (cannot link to a refund) → 404.
+
+    This is a genuine "no refund will ever exist" case, distinct from the
+    pending 425 path.
+    """
+    from fastapi import HTTPException
+
+    x_cashu_token = "cashuAnoreqid_token"
+    in_tx = _make_cashu_tx(
+        token=x_cashu_token, amount=0, unit="msat", type="in", request_id=None
+    )
+
+    session = MagicMock()
+    session.exec = AsyncMock(side_effect=[_exec_result(in_tx)])
+
+    with pytest.raises(HTTPException) as exc_info:
+        await refund_wallet_endpoint(
+            authorization="Bearer sk-somekey",
+            x_cashu=x_cashu_token,
+            session=session,
+        )
+
+    assert exc_info.value.status_code == 404
+
+
 @pytest.mark.asyncio
 async def test_refund_x_cashu_swept_raises_410() -> None:
     from fastapi import HTTPException
@@ -339,7 +398,10 @@ async def test_apikey_refund_restores_balance_on_mint_failure() -> None:
 
     with (
         patch("routstr.balance.get_billing_key", AsyncMock(return_value=key)),
-        patch("routstr.balance.send_token", AsyncMock(side_effect=Exception("mint down"))),
+        patch(
+            "routstr.balance.send_token",
+            AsyncMock(side_effect=MintConnectionError("raw mint outage detail")),
+        ),
         patch("routstr.balance.store_cashu_transaction", AsyncMock()),
         patch("routstr.balance._refund_cache_get", AsyncMock(return_value=None)),
         patch("routstr.balance._refund_cache_set", AsyncMock()),
@@ -353,10 +415,46 @@ async def test_apikey_refund_restores_balance_on_mint_failure() -> None:
             )
 
     assert exc_info.value.status_code == 503
+    assert exc_info.value.detail == "Mint service unavailable"
+    assert "raw mint outage detail" not in exc_info.value.detail
     # Verify two exec calls: debit + restore
     assert session.exec.await_count == 2
 
 
+@pytest.mark.asyncio
+async def test_apikey_refund_generic_failure_is_sanitized_500() -> None:
+    """Unexpected send-side failures restore balance without leaking exception text."""
+    from fastapi import HTTPException
+
+    key = _make_api_key(balance=5000, refund_currency="sat")
+    raw_error = "database secret token raw-mint-response"
+
+    session = MagicMock()
+    session.get = AsyncMock(return_value=key)
+    session.exec = AsyncMock(side_effect=[_update_result(1), _update_result(1)])
+    session.commit = AsyncMock()
+
+    with (
+        patch("routstr.balance.get_billing_key", AsyncMock(return_value=key)),
+        patch("routstr.balance.send_token", AsyncMock(side_effect=RuntimeError(raw_error))),
+        patch("routstr.balance.store_cashu_transaction", AsyncMock()),
+        patch("routstr.balance._refund_cache_get", AsyncMock(return_value=None)),
+        patch("routstr.balance._refund_cache_set", AsyncMock()),
+        patch("routstr.balance.logger"),
+    ):
+        with pytest.raises(HTTPException) as exc_info:
+            await refund_wallet_endpoint(
+                authorization="Bearer sk-testhash",
+                x_cashu=None,
+                session=session,
+            )
+
+    assert exc_info.value.status_code == 500
+    assert exc_info.value.detail == "Refund failed"
+    assert raw_error not in exc_info.value.detail
+    assert session.exec.await_count == 2
+
+
 # ---------------------------------------------------------------------------
 # no-create guarantee: fresh Cashu/unknown sk- tokens must not create API keys
 # ---------------------------------------------------------------------------
@@ -401,3 +499,190 @@ async def test_refund_unknown_sk_bearer_returns_401() -> None:
 
     assert exc_info.value.status_code == 401
     session.get.assert_awaited_once()
+
+
+# --- Topup redemption error taxonomy (POST /v1/wallet/topup) ------------------
+
+
+@pytest.mark.asyncio
+@pytest.mark.parametrize(
+    "error",
+    [
+        httpx.ConnectError("All connection attempts failed"),
+        MintConnectionError("connect to mint refused"),
+        TimeoutError("timed out connecting to mint"),
+    ],
+)
+async def test_topup_mint_unreachable_returns_503(error: Exception) -> None:
+    """A down mint must surface 503 (retryable), not 400 or 500 — the token is
+    fine, so the client should retry once the mint recovers."""
+    from fastapi import HTTPException
+
+    key = _make_api_key(balance=1000)
+    session = MagicMock()
+
+    with (
+        patch("routstr.balance.get_billing_key", AsyncMock(return_value=key)),
+        patch("routstr.balance.credit_balance", AsyncMock(side_effect=error)),
+    ):
+        with pytest.raises(HTTPException) as exc_info:
+            await topup_wallet_endpoint(
+                cashu_token="cashuAtoken", key=key, session=session
+            )
+
+    assert exc_info.value.status_code == 503
+    assert exc_info.value.detail == "Cashu mint is unreachable"
+
+
+@pytest.mark.asyncio
+async def test_topup_already_spent_still_returns_400() -> None:
+    """Regression: the mint-unreachable short-circuit must not swallow the
+    existing ValueError substring buckets."""
+    from fastapi import HTTPException
+
+    key = _make_api_key(balance=1000)
+    session = MagicMock()
+
+    with (
+        patch("routstr.balance.get_billing_key", AsyncMock(return_value=key)),
+        patch(
+            "routstr.balance.credit_balance",
+            AsyncMock(side_effect=ValueError("Token already spent")),
+        ),
+    ):
+        with pytest.raises(HTTPException) as exc_info:
+            await topup_wallet_endpoint(
+                cashu_token="cashuAtoken", key=key, session=session
+            )
+
+    assert exc_info.value.status_code == 400
+    assert exc_info.value.detail == "Cashu token already spent"
+
+
+@pytest.mark.asyncio
+async def test_topup_zero_value_returns_400_zero_value_message() -> None:
+    """A dust/zero redemption maps to the documented zero-value message, not the
+    generic redemption-failed one."""
+    from fastapi import HTTPException
+
+    key = _make_api_key(balance=1000)
+    session = MagicMock()
+
+    with (
+        patch("routstr.balance.get_billing_key", AsyncMock(return_value=key)),
+        patch(
+            "routstr.balance.credit_balance",
+            AsyncMock(
+                side_effect=ValueError("Redeemed token amount must be positive, got 0 msats")
+            ),
+        ),
+    ):
+        with pytest.raises(HTTPException) as exc_info:
+            await topup_wallet_endpoint(
+                cashu_token="cashuAtoken", key=key, session=session
+            )
+
+    assert exc_info.value.status_code == 400
+    assert exc_info.value.detail == "Failed to redeem Cashu token: token yielded no value"
+
+
+@pytest.mark.asyncio
+async def test_topup_token_consumed_returns_500() -> None:
+    """A post-redemption crediting failure (token spent) is a non-retryable 500,
+    not a 4xx that invites a retry."""
+    from fastapi import HTTPException
+
+    from routstr.wallet import TokenConsumedError
+
+    key = _make_api_key(balance=1000)
+    session = MagicMock()
+
+    with (
+        patch("routstr.balance.get_billing_key", AsyncMock(return_value=key)),
+        patch(
+            "routstr.balance.credit_balance",
+            AsyncMock(side_effect=TokenConsumedError("credit failed")),
+        ),
+    ):
+        with pytest.raises(HTTPException) as exc_info:
+            await topup_wallet_endpoint(
+                cashu_token="cashuAtoken", key=key, session=session
+            )
+
+    assert exc_info.value.status_code == 500
+    assert exc_info.value.detail == (
+        "Token was redeemed but could not be credited; do not retry"
+    )
+
+
+@pytest.mark.asyncio
+@pytest.mark.parametrize(
+    ("error", "expected_status", "expected_detail"),
+    [
+        (
+            ValueError(
+                "Failed to estimate fees: Fees (7 sat) exceed token amount (5 sat)"
+            ),
+            422,
+            "Token value is too small to cover swap fees",
+        ),
+        (
+            ValueError(
+                "Token amount (5 sat) is insufficient to cover melt fees."
+            ),
+            422,
+            "Token value is too small to cover swap fees",
+        ),
+        (
+            ValueError("Failed to melt token from foreign mint http://m: boom"),
+            422,
+            "Failed to swap token from foreign mint",
+        ),
+    ],
+)
+async def test_topup_fee_and_swap_failures_return_422(
+    error: Exception, expected_status: int, expected_detail: str
+) -> None:
+    """Fee/swap failures map to 422 (shared taxonomy), matching the bearer and
+    X-Cashu paths — previously top-up flattened these to 400."""
+    from fastapi import HTTPException
+
+    key = _make_api_key(balance=1000)
+    session = MagicMock()
+
+    with (
+        patch("routstr.balance.get_billing_key", AsyncMock(return_value=key)),
+        patch("routstr.balance.credit_balance", AsyncMock(side_effect=error)),
+    ):
+        with pytest.raises(HTTPException) as exc_info:
+            await topup_wallet_endpoint(
+                cashu_token="cashuAtoken", key=key, session=session
+            )
+
+    assert exc_info.value.status_code == expected_status
+    assert exc_info.value.detail == expected_detail
+
+
+@pytest.mark.asyncio
+async def test_topup_unexpected_non_valueerror_returns_500() -> None:
+    """A non-ValueError, non-transport fault is an internal error (500), not a
+    sanitized 400 — the merged except must preserve this."""
+    from fastapi import HTTPException
+
+    key = _make_api_key(balance=1000)
+    session = MagicMock()
+
+    with (
+        patch("routstr.balance.get_billing_key", AsyncMock(return_value=key)),
+        patch(
+            "routstr.balance.credit_balance",
+            AsyncMock(side_effect=RuntimeError("db exploded")),
+        ),
+    ):
+        with pytest.raises(HTTPException) as exc_info:
+            await topup_wallet_endpoint(
+                cashu_token="cashuAtoken", key=key, session=session
+            )
+
+    assert exc_info.value.status_code == 500
+    assert exc_info.value.detail == "Internal server error"

tests/unit/test_balance_create_api.py

@@ -0,0 +1,40 @@
+from types import SimpleNamespace
+from unittest.mock import AsyncMock
+
+import pytest
+from fastapi import FastAPI
+from httpx import ASGITransport, AsyncClient
+
+from routstr import balance as balance_module
+from routstr.core.db import get_session
+
+
+@pytest.mark.asyncio
+async def test_create_balance_accepts_large_cashu_token_in_post_body(
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    token = "cashuA" + "x" * 20_000
+    key = SimpleNamespace(hashed_key="hashed", balance=123_000)
+    validate_bearer_key = AsyncMock(return_value=key)
+    session = AsyncMock()
+    monkeypatch.setattr(balance_module, "validate_bearer_key", validate_bearer_key)
+
+    async def override_get_session():  # type: ignore[no-untyped-def]
+        yield session
+
+    app = FastAPI()
+    app.include_router(balance_module.balance_router)
+    app.dependency_overrides[get_session] = override_get_session
+
+    async with AsyncClient(
+        transport=ASGITransport(app=app),  # type: ignore[arg-type]
+        base_url="http://test",
+    ) as client:
+        response = await client.post(
+            "/v1/balance/create",
+            json={"initial_balance_token": token},
+        )
+
+    assert response.status_code == 200
+    assert response.json() == {"api_key": "sk-hashed", "balance": 123_000}
+    validate_bearer_key.assert_awaited_once_with(token, session)

tests/unit/test_balances_by_mint_and_unit.py

@@ -0,0 +1,115 @@
+"""Real-DB coverage for db.balances_by_mint_and_unit.
+
+Verifies the grouped liability query used by fetch_all_balances: it sums
+balances per (mint_url, unit), filters to the requested mints/units, excludes
+NULL mint/currency rows, and returns nothing for empty inputs.
+"""
+
+from typing import AsyncGenerator
+
+import pytest
+from sqlalchemy.ext.asyncio import AsyncEngine, create_async_engine
+from sqlalchemy.pool import StaticPool
+from sqlmodel import SQLModel
+from sqlmodel.ext.asyncio.session import AsyncSession
+
+from routstr.core.db import (
+    ApiKey,
+    balance_for_mint_and_unit,
+    balances_by_mint_and_unit,
+)
+
+
+def _make_engine() -> AsyncEngine:
+    return create_async_engine(
+        "sqlite+aiosqlite://",
+        poolclass=StaticPool,
+        connect_args={"check_same_thread": False},
+    )
+
+
+@pytest.fixture
+async def session() -> "AsyncGenerator[AsyncSession, None]":
+    engine = _make_engine()
+    async with engine.begin() as conn:
+        await conn.run_sync(SQLModel.metadata.create_all)
+    db_session = AsyncSession(engine, expire_on_commit=False)
+    try:
+        yield db_session
+    finally:
+        await db_session.close()
+        await engine.dispose()
+
+
+async def _add_key(
+    session: AsyncSession,
+    hashed_key: str,
+    balance: int,
+    mint_url: str | None,
+    currency: str | None,
+) -> None:
+    session.add(
+        ApiKey(
+            hashed_key=hashed_key,
+            balance=balance,
+            refund_mint_url=mint_url,
+            refund_currency=currency,
+        )
+    )
+    await session.commit()
+
+
+@pytest.mark.asyncio
+async def test_sums_and_groups_by_mint_and_unit(session: AsyncSession) -> None:
+    await _add_key(session, "a", 1000, "http://m1", "sat")
+    await _add_key(session, "b", 500, "http://m1", "sat")
+    await _add_key(session, "c", 7000, "http://m1", "msat")
+    await _add_key(session, "d", 200, "http://m2", "sat")
+
+    result = await balances_by_mint_and_unit(
+        session, ["http://m1", "http://m2"], ["sat", "msat"]
+    )
+
+    assert result[("http://m1", "sat")] == 1500
+    assert result[("http://m1", "msat")] == 7000
+    assert result[("http://m2", "sat")] == 200
+
+
+@pytest.mark.asyncio
+async def test_filters_out_unrequested_mints_and_units(session: AsyncSession) -> None:
+    await _add_key(session, "a", 1000, "http://wanted", "sat")
+    await _add_key(session, "b", 999, "http://other", "sat")
+    await _add_key(session, "c", 888, "http://wanted", "usd")
+
+    result = await balances_by_mint_and_unit(session, ["http://wanted"], ["sat"])
+
+    assert result == {("http://wanted", "sat"): 1000}
+
+
+@pytest.mark.asyncio
+async def test_excludes_rows_with_null_mint_or_currency(session: AsyncSession) -> None:
+    await _add_key(session, "a", 1000, "http://m1", "sat")
+    await _add_key(session, "b", 4242, None, None)
+
+    result = await balances_by_mint_and_unit(session, ["http://m1"], ["sat"])
+
+    assert result == {("http://m1", "sat"): 1000}
+
+
+@pytest.mark.asyncio
+async def test_scalar_balance_for_one_mint_and_unit(session: AsyncSession) -> None:
+    await _add_key(session, "a", 1000, "http://m1", "sat")
+    await _add_key(session, "b", 500, "http://m1", "sat")
+    await _add_key(session, "c", 9000, "http://m1", "msat")
+    await _add_key(session, "d", 700, "http://m2", "sat")
+
+    assert await balance_for_mint_and_unit(session, "http://m1", "sat") == 1500
+    assert await balance_for_mint_and_unit(session, "http://missing", "sat") == 0
+
+
+@pytest.mark.asyncio
+async def test_empty_inputs_return_empty_mapping(session: AsyncSession) -> None:
+    await _add_key(session, "a", 1000, "http://m1", "sat")
+
+    assert await balances_by_mint_and_unit(session, [], ["sat"]) == {}
+    assert await balances_by_mint_and_unit(session, ["http://m1"], []) == {}

tests/unit/test_cache_breakpoints.py

@@ -0,0 +1,189 @@
+"""Tests for Anthropic cache-breakpoint injection on forwarded requests.
+
+Anthropic prompt caching is explicit; a client that doesn't recognise a routstr
+URL as Anthropic-backed never sends ``cache_control`` markers, so caching never
+engages over routstr. ``prepare_request_body`` must stamp the standard
+breakpoints for Anthropic-family models while always deferring to client-set
+markers and never touching automatic-cache providers.
+"""
+
+import json
+import os
+
+import pytest
+
+os.environ.setdefault("UPSTREAM_BASE_URL", "http://test")
+os.environ.setdefault("UPSTREAM_API_KEY", "test")
+os.environ.setdefault("LIGHTNING_ADDRESS", "test@stm.to")
+
+from routstr.upstream import GenericUpstreamProvider
+from routstr.upstream.cache_breakpoints import (
+    body_has_cache_control,
+    inject_anthropic_cache_breakpoints,
+    is_explicit_cache_model,
+)
+
+
+def _chat_body() -> dict:
+    return {
+        "model": "anthropic/claude-sonnet-4.5",
+        "stream": True,
+        "messages": [
+            {"role": "system", "content": "You are concise."},
+            {"role": "user", "content": "Hello"},
+        ],
+        "tools": [
+            {"type": "function", "function": {"name": "a"}},
+            {"type": "function", "function": {"name": "b"}},
+        ],
+    }
+
+
+# ---------------------------------------------------------------------------
+# Model detection
+# ---------------------------------------------------------------------------
+
+
+@pytest.mark.parametrize(
+    "model_id,expected",
+    [
+        ("anthropic/claude-sonnet-4.5", True),
+        ("claude-haiku-4-5-20251001", True),
+        # Alibaba explicit-cache models share Anthropic's wire format
+        ("qwen/qwen3-max", True),
+        ("qwen/qwen3-coder-plus", True),
+        ("deepseek/deepseek-v3.2", True),
+        # Automatic-cache providers need no markers
+        ("openai/gpt-4o", False),
+        ("google/gemini-2.5-flash", False),
+        ("deepseek/deepseek-chat", False),
+        ("qwen/qwen3.5-plus-02-15", False),  # snapshot, no explicit caching
+        (None, False),
+    ],
+)
+def test_is_explicit_cache_model(model_id: str | None, expected: bool) -> None:
+    assert is_explicit_cache_model(model_id) is expected
+
+
+def test_is_explicit_cache_model_uses_fallbacks() -> None:
+    # routstr id is opaque but a forwarded/canonical alias reveals the family.
+    assert is_explicit_cache_model("model-xyz", None, "anthropic/claude-opus-4.1")
+
+
+# ---------------------------------------------------------------------------
+# Breakpoint placement
+# ---------------------------------------------------------------------------
+
+
+def test_injects_three_breakpoints() -> None:
+    data = _chat_body()
+    assert inject_anthropic_cache_breakpoints(data) is True
+
+    # system prompt promoted to array form with a marker
+    system = data["messages"][0]["content"]
+    assert system == [
+        {
+            "type": "text",
+            "text": "You are concise.",
+            "cache_control": {"type": "ephemeral"},
+        }
+    ]
+    # last tool marked
+    assert data["tools"][-1]["cache_control"] == {"type": "ephemeral"}
+    assert "cache_control" not in data["tools"][0]
+    # last user message marked
+    user = data["messages"][1]["content"]
+    assert user[-1]["cache_control"] == {"type": "ephemeral"}
+
+
+def test_defers_to_client_supplied_cache_control() -> None:
+    data = _chat_body()
+    data["messages"][1]["content"] = [
+        {"type": "text", "text": "Hello", "cache_control": {"type": "ephemeral"}}
+    ]
+    assert body_has_cache_control(data) is True
+    # No additional stamping when the client already controls caching.
+    assert inject_anthropic_cache_breakpoints(data) is False
+    assert "cache_control" not in data["tools"][-1]
+
+
+def test_marks_last_text_part_of_array_content() -> None:
+    data = _chat_body()
+    data["messages"][1]["content"] = [
+        {"type": "text", "text": "first"},
+        {"type": "image_url", "image_url": {"url": "x"}},
+        {"type": "text", "text": "last"},
+    ]
+    inject_anthropic_cache_breakpoints(data)
+    parts = data["messages"][1]["content"]
+    assert parts[2]["cache_control"] == {"type": "ephemeral"}
+    assert "cache_control" not in parts[0]
+
+
+def test_noop_without_messages() -> None:
+    assert inject_anthropic_cache_breakpoints({"prompt": "x"}) is False
+
+
+# ---------------------------------------------------------------------------
+# prepare_request_body integration
+# ---------------------------------------------------------------------------
+
+
+def _model(model_id: str):  # type: ignore[no-untyped-def]
+    from routstr.payment.models import Architecture, Model, Pricing
+
+    return Model(
+        id=model_id,
+        name=model_id,
+        created=0,
+        description="",
+        context_length=200000,
+        architecture=Architecture(
+            modality="text->text",
+            input_modalities=["text"],
+            output_modalities=["text"],
+            tokenizer="Claude",
+            instruct_type=None,
+        ),
+        pricing=Pricing(prompt=0.0, completion=0.0),
+    )
+
+
+def _openrouter_provider() -> "GenericUpstreamProvider":
+    # OpenRouter endpoint via the generic provider — recognised by base URL.
+    return GenericUpstreamProvider(base_url="https://openrouter.ai/api/v1")
+
+
+@pytest.mark.parametrize(
+    "model_id", ["anthropic/claude-sonnet-4.5", "qwen/qwen3-max", "deepseek/deepseek-v3.2"]
+)
+def test_prepare_request_body_injects_for_explicit_models(model_id: str) -> None:
+    provider = _openrouter_provider()
+    body = json.dumps(_chat_body()).encode()
+    out = provider.prepare_request_body(body, _model(model_id))
+    assert out is not None
+    data = json.loads(out)
+    assert body_has_cache_control(data) is True
+    assert data["tools"][-1]["cache_control"] == {"type": "ephemeral"}
+
+
+def test_prepare_request_body_skips_for_automatic_provider_model() -> None:
+    provider = _openrouter_provider()
+    body = json.dumps(_chat_body()).encode()
+    out = provider.prepare_request_body(body, _model("openai/gpt-4o"))
+    assert out is not None
+    data = json.loads(out)
+    assert body_has_cache_control(data) is False
+
+
+def test_prepare_request_body_skips_when_upstream_rejects_markers() -> None:
+    # Claude id but a non-OpenRouter/Anthropic upstream → must NOT inject,
+    # since the markers could be rejected by an upstream that doesn't accept them.
+    from routstr.upstream import GenericUpstreamProvider
+
+    provider = GenericUpstreamProvider(base_url="https://some-gateway.example/v1")
+    body = json.dumps(_chat_body()).encode()
+    out = provider.prepare_request_body(body, _model("anthropic/claude-sonnet-4.5"))
+    assert out is not None
+    data = json.loads(out)
+    assert body_has_cache_control(data) is False

tests/unit/test_cache_pricing.py

@@ -0,0 +1,347 @@
+"""Tests for cache-aware pricing of cached input tokens.
+
+Specifies two things:
+
+1. ``backfill_cache_pricing`` — when the OpenRouter model feed omits cache
+   rates (it does for most DeepSeek models and e.g. openai/gpt-4o), they are
+   filled from litellm's bundled cost map instead of silently billing cache
+   reads at the full input rate. Existing OpenRouter values are never
+   overwritten, and provider fees apply to backfilled rates like any other.
+2. ``calculate_cost`` — cached tokens are billed at the cache rates from the
+   model's sats_pricing; the full input rate remains only as the documented
+   last resort when no cache rate could be resolved anywhere.
+"""
+
+import os
+from unittest.mock import Mock, patch
+
+import litellm
+import pytest
+
+os.environ.setdefault("UPSTREAM_BASE_URL", "http://test")
+os.environ.setdefault("UPSTREAM_API_KEY", "test")
+os.environ.setdefault("LIGHTNING_ADDRESS", "test@stm.to")
+
+from routstr.core.settings import settings
+from routstr.payment.cost_calculation import CostData, calculate_cost
+from routstr.payment.models import (
+    Architecture,
+    Model,
+    Pricing,
+    backfill_cache_pricing,
+)
+from routstr.upstream import GenericUpstreamProvider
+
+
+def _make_model(model_id: str, pricing: Pricing) -> Model:
+    return Model(
+        id=model_id,
+        name=model_id,
+        created=0,
+        description="",
+        context_length=64000,
+        architecture=Architecture(
+            modality="text->text",
+            input_modalities=["text"],
+            output_modalities=["text"],
+            tokenizer="Other",
+            instruct_type=None,
+        ),
+        pricing=pricing,
+    )
+
+
+# ============================================================================
+# backfill_cache_pricing — litellm as fallback source for missing cache rates
+# ============================================================================
+
+
+def test_backfill_deepseek_cache_read_from_litellm() -> None:
+    """deepseek/deepseek-chat has no input_cache_read on OpenRouter; litellm
+    knows the real rate (10x cheaper than input)."""
+    pricing = Pricing(prompt=2.8e-07, completion=4.2e-07)
+
+    result = backfill_cache_pricing("deepseek/deepseek-chat", pricing)
+
+    expected = litellm.model_cost["deepseek/deepseek-chat"][
+        "cache_read_input_token_cost"
+    ]
+    assert result.input_cache_read == expected
+    assert result.input_cache_read < pricing.prompt  # sanity: it's a discount
+
+
+def test_backfill_strips_vendor_prefix_for_litellm_lookup() -> None:
+    """OpenRouter ids are vendor-prefixed (openai/gpt-4o); litellm keys most
+    non-DeepSeek models without the prefix (gpt-4o)."""
+    pricing = Pricing(prompt=2.5e-06, completion=1e-05)
+
+    result = backfill_cache_pricing("openai/gpt-4o", pricing)
+
+    expected = litellm.model_cost["gpt-4o"]["cache_read_input_token_cost"]
+    assert result.input_cache_read == expected
+
+
+def test_backfill_case_insensitive_lookup() -> None:
+    """A generic upstream may report a mixed-case id
+    (deepseek-ai/DeepSeek-V4-Flash); litellm keys are lowercase. The
+    case-insensitive fallback still resolves the cache rate."""
+    pricing = Pricing(prompt=1.4e-07, completion=2.8e-07)
+
+    result = backfill_cache_pricing("deepseek-ai/DeepSeek-V4-Flash", pricing)
+
+    expected = litellm.model_cost["deepseek-v4-flash"][
+        "cache_read_input_token_cost"
+    ]
+    assert result.input_cache_read == expected
+    assert result.input_cache_read < pricing.prompt  # sanity: it's a discount
+
+
+def test_backfill_fills_cache_write_rate() -> None:
+    """Anthropic cache writes cost more than input (1.25x); billing them at
+    the input rate undercharges. litellm carries the write rate."""
+    pricing = Pricing(prompt=3e-06, completion=1.5e-05)
+
+    result = backfill_cache_pricing("anthropic/claude-sonnet-4-5", pricing)
+
+    expected = litellm.model_cost["claude-sonnet-4-5"][
+        "cache_creation_input_token_cost"
+    ]
+    assert result.input_cache_write == expected
+    assert result.input_cache_write > pricing.prompt  # sanity: write premium
+
+
+def test_backfill_never_overwrites_openrouter_rates() -> None:
+    """When OpenRouter provides a cache rate, it is authoritative."""
+    pricing = Pricing(
+        prompt=2.1e-07, completion=7.9e-07, input_cache_read=1.3e-07
+    )
+
+    result = backfill_cache_pricing("deepseek/deepseek-chat", pricing)
+
+    assert result.input_cache_read == 1.3e-07
+
+
+def test_backfill_unknown_model_unchanged() -> None:
+    """Models litellm doesn't know stay untouched (last-resort fallback to
+    the input rate happens later, at billing time)."""
+    pricing = Pricing(prompt=1e-06, completion=2e-06)
+
+    result = backfill_cache_pricing("artificial-dumbness/dumb-1", pricing)
+
+    assert result.input_cache_read == 0.0
+    assert result.input_cache_write == 0.0
+
+
+def test_provider_fee_applies_to_backfilled_cache_rates() -> None:
+    """Backfill happens before the provider fee, so cache rates carry the
+    same markup as every other price component."""
+    provider = GenericUpstreamProvider(
+        base_url="http://upstream.example", provider_fee=2.0
+    )
+    model = _make_model(
+        "deepseek/deepseek-chat", Pricing(prompt=2.8e-07, completion=4.2e-07)
+    )
+
+    adjusted = provider._apply_provider_fee_to_model(model)
+
+    litellm_read = litellm.model_cost["deepseek/deepseek-chat"][
+        "cache_read_input_token_cost"
+    ]
+    assert adjusted.pricing.input_cache_read == pytest.approx(litellm_read * 2.0)
+    assert adjusted.pricing.prompt == pytest.approx(2.8e-07 * 2.0)
+
+
+def test_row_to_model_backfills_cache_rate() -> None:
+    """The DB-override path (admin-configured providers, e.g. a generic
+    upstream) stores pricing without cache rates. ``_row_to_model`` must
+    backfill them from litellm just like ``_apply_provider_fee_to_model``,
+    otherwise cache reads bill at the full input rate."""
+    import json
+
+    from routstr.core.db import ModelRow
+    from routstr.payment.models import _row_to_model
+
+    row = ModelRow(
+        id="deepseek-v4-flash",
+        name="deepseek-v4-flash",
+        created=0,
+        description="",
+        context_length=1000000,
+        architecture=json.dumps(
+            {
+                "modality": "text",
+                "input_modalities": ["text"],
+                "output_modalities": ["text"],
+                "tokenizer": "unknown",
+                "instruct_type": None,
+            }
+        ),
+        # Stored pricing omits input_cache_read (generic provider never sets it).
+        pricing=json.dumps({"prompt": 1.4e-07, "completion": 2.8e-07}),
+        enabled=True,
+        upstream_provider_id=1,
+    )
+
+    with patch(
+        "routstr.payment.models.sats_usd_price", return_value=5.0e-5
+    ):
+        model = _row_to_model(row, apply_provider_fee=True, provider_fee=1.0)
+
+    litellm_read = litellm.model_cost["deepseek-v4-flash"][
+        "cache_read_input_token_cost"
+    ]
+    assert model.pricing.input_cache_read == pytest.approx(litellm_read)
+    assert model.pricing.input_cache_read < model.pricing.prompt  # a discount
+    assert model.sats_pricing is not None
+    assert model.sats_pricing.input_cache_read > 0
+
+
+def test_row_to_model_backfills_via_forwarded_model_id() -> None:
+    """An alias row (id != forwarded_model_id) must backfill cache rates from
+    the *forwarded* model name — the real upstream model litellm prices —
+    not the alias id, which litellm doesn't know."""
+    import json
+
+    from routstr.core.db import ModelRow
+    from routstr.payment.models import _row_to_model
+
+    row = ModelRow(
+        id="local-alias",  # litellm has no such key
+        name="local-alias",
+        created=0,
+        description="",
+        context_length=1000000,
+        architecture=json.dumps(
+            {
+                "modality": "text",
+                "input_modalities": ["text"],
+                "output_modalities": ["text"],
+                "tokenizer": "unknown",
+                "instruct_type": None,
+            }
+        ),
+        pricing=json.dumps({"prompt": 1.4e-07, "completion": 2.8e-07}),
+        enabled=True,
+        upstream_provider_id=1,
+        forwarded_model_id="deepseek-v4-flash",
+    )
+
+    with patch(
+        "routstr.payment.models.sats_usd_price", return_value=5.0e-5
+    ):
+        model = _row_to_model(row, apply_provider_fee=True, provider_fee=1.0)
+
+    litellm_read = litellm.model_cost["deepseek-v4-flash"][
+        "cache_read_input_token_cost"
+    ]
+    assert model.pricing.input_cache_read == pytest.approx(litellm_read)
+    assert model.pricing.input_cache_read < model.pricing.prompt
+
+
+# ============================================================================
+# calculate_cost — cached tokens billed at cache rates
+# ============================================================================
+
+
+@pytest.fixture(autouse=True)
+def patch_sats_usd_price() -> None:  # type: ignore[misc]
+    with patch("routstr.payment.cost_calculation.sats_usd_price", return_value=5.0e-5):
+        yield
+
+
+@pytest.fixture
+def model_pricing(monkeypatch: pytest.MonkeyPatch) -> Mock:
+    """Model-based pricing: 1 msat per input token, 2 per output token,
+    0.1 per cache-read token, 1.25 per cache-write token."""
+    monkeypatch.setattr(settings, "fixed_pricing", False)
+    model = Mock()
+    model.sats_pricing = Pricing(
+        prompt=0.001,
+        completion=0.002,
+        input_cache_read=0.0001,
+        input_cache_write=0.00125,
+    )
+    return model
+
+
+@pytest.mark.asyncio
+async def test_deepseek_cache_hits_billed_at_cache_rate(model_pricing: Mock) -> None:
+    """The reported overcharge scenario: a 10k-token prompt with 90% cache
+    hits costs 2900 msats at honest rates, not the 11000 msats that billing
+    every prompt token at the full input rate would charge."""
+    response = {
+        "model": "deepseek-chat",
+        "usage": {
+            "prompt_tokens": 10000,
+            "completion_tokens": 500,
+            "prompt_cache_hit_tokens": 9000,
+            "prompt_cache_miss_tokens": 1000,
+        },
+    }
+
+    with patch("routstr.proxy.get_model_instance", return_value=model_pricing):
+        result = await calculate_cost(response, max_cost=100000)
+
+    assert isinstance(result, CostData)
+    # 1000 input @ 1 msat + 9000 cache reads @ 0.1 msat + 500 output @ 2 msat.
+    # input_msats folds the cache-read cost in (1000 + 900) so a dashboard
+    # rendering I/O/T sees input + output == total; the cache portion stays
+    # visible in cache_read_msats.
+    assert result.cache_read_msats == 900
+    assert result.output_msats == 1000
+    assert result.input_msats == 1900
+    assert result.input_msats + result.output_msats == result.total_msats
+    assert result.total_msats == 2900
+
+
+@pytest.mark.asyncio
+async def test_anthropic_cache_write_billed_at_write_rate(model_pricing: Mock) -> None:
+    """Cache writes carry their premium rate (1.25x input here), instead of
+    being silently billed at the plain input rate."""
+    response = {
+        "model": "claude-sonnet-4-5",
+        "usage": {
+            "input_tokens": 300,
+            "output_tokens": 100,
+            "cache_read_input_tokens": 500,
+            "cache_creation_input_tokens": 2000,
+        },
+    }
+
+    with patch("routstr.proxy.get_model_instance", return_value=model_pricing):
+        result = await calculate_cost(response, max_cost=100000)
+
+    assert isinstance(result, CostData)
+    # 300 @ 1 + 500 @ 0.1 + 2000 @ 1.25 + 100 @ 2
+    assert result.cache_read_msats == 50
+    assert result.cache_creation_msats == 2500
+    assert result.total_msats == 3050
+
+
+@pytest.mark.asyncio
+async def test_missing_cache_rate_falls_back_to_input_rate(
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    """Documented last resort: when no cache rate could be resolved anywhere
+    (OpenRouter and litellm both silent), cache reads bill at the input rate —
+    never cheaper, never free."""
+    monkeypatch.setattr(settings, "fixed_pricing", False)
+    model = Mock()
+    model.sats_pricing = Pricing(prompt=0.001, completion=0.002)
+
+    response = {
+        "model": "dumb-1",
+        "usage": {
+            "prompt_tokens": 10000,
+            "completion_tokens": 500,
+            "prompt_cache_hit_tokens": 9000,
+            "prompt_cache_miss_tokens": 1000,
+        },
+    }
+
+    with patch("routstr.proxy.get_model_instance", return_value=model):
+        result = await calculate_cost(response, max_cost=100000)
+
+    assert isinstance(result, CostData)
+    # 1000 @ 1 + 9000 @ 1 (fallback) + 500 @ 2
+    assert result.total_msats == 11000