diff --git a/routstr/upstream/generic.py b/routstr/upstream/generic.py
index 3faf80e9..63b6ffc5 100644
--- a/routstr/upstream/generic.py
+++ b/routstr/upstream/generic.py
@@ -91,6 +91,19 @@ class GenericUpstreamProvider(BaseUpstreamProvider):
         if input_usd < 0 or output_usd < 0 or (input_usd == 0 and output_usd == 0):
             return None
 
+        # Venice ships a discounted cache-read rate as ``cache_input`` (e.g.
+        # deepseek-v4-1-flash: $0.0075/1M vs $0.375/1M input). Dropping it
+        # left ``input_cache_read`` at 0, which billing reads as "no cache
+        # rate" and falls back to the FULL input rate — a 50x overcharge on
+        # cache hits. A malformed/negative cache rate coerces to None and is
+        # treated as absent, never carried (same rule as the OpenRouter rung).
+        cache_read_usd = _as_float(pricing_info.get("cache_input", {}).get("usd"))
+        input_cache_read = (
+            cache_read_usd / 1_000_000
+            if cache_read_usd is not None and cache_read_usd > 0
+            else 0.0
+        )
+
         capabilities = model_spec.get("capabilities", {})
         input_modalities = ["text"]
         if capabilities.get("supportsVision", False):
@@ -101,6 +114,7 @@ class GenericUpstreamProvider(BaseUpstreamProvider):
             completion=output_usd / 1_000_000,
             context_length=model_spec.get("availableContextTokens"),
             source="native",
+            input_cache_read=input_cache_read,
             input_modalities=input_modalities,
         )
 
diff --git a/tests/unit/test_upstream_generic.py b/tests/unit/test_upstream_generic.py
index 2e6fd56e..175aa15d 100644
--- a/tests/unit/test_upstream_generic.py
+++ b/tests/unit/test_upstream_generic.py
@@ -111,6 +111,85 @@ async def test_native_model_spec_resolves_and_captures_metadata() -> None:
     or_feed.assert_not_awaited()
 
 
+@pytest.mark.asyncio
+async def test_native_model_spec_carries_cache_read_price() -> None:
+    """Venice's ``cache_input`` rate must reach ``input_cache_read``.
+
+    Regression: the native parser read only ``input``/``output`` and dropped
+    ``cache_input``, leaving ``input_cache_read`` at 0 — which billing reads
+    as "no cache rate" and falls back to the FULL input rate, a ~50x
+    overcharge on cache hits (deepseek-v4-1-flash: $0.0075 vs $0.375 per 1M).
+    """
+    payload = {
+        "data": [
+            {
+                "id": "venice-flash",
+                "owned_by": "venice",
+                "model_spec": {
+                    "name": "Venice Flash",
+                    "availableContextTokens": 1000000,
+                    "pricing": {
+                        "input": {"usd": 0.375},
+                        "cache_input": {"usd": 0.0075},
+                        "output": {"usd": 1.5},
+                    },
+                },
+            }
+        ]
+    }
+
+    with _patch_models_endpoint(payload):
+        with patch(
+            "routstr.payment.models.async_fetch_openrouter_models",
+            AsyncMock(return_value=[]),
+        ):
+            models = await GenericUpstreamProvider(base_url="http://x").fetch_models()
+
+    model = _model_by_id(models, "venice-flash")
+    assert model.enabled is True
+    assert model.pricing.prompt == pytest.approx(0.375 / 1_000_000)
+    assert model.pricing.input_cache_read == pytest.approx(0.0075 / 1_000_000)
+
+
+@pytest.mark.asyncio
+@pytest.mark.parametrize("bad_rate", [-0.0075, "Infinity", "not-a-number"])
+async def test_native_model_spec_drops_malformed_cache_rate(bad_rate: object) -> None:
+    """A malformed ``cache_input`` costs the cache rate, not the model.
+
+    The token prices still resolve; the cache rate falls back to 0 (absent),
+    mirroring the OpenRouter rung's drop-don't-carry rule.
+    """
+    payload = {
+        "data": [
+            {
+                "id": "venice-badcache",
+                "owned_by": "venice",
+                "model_spec": {
+                    "name": "Venice Bad Cache",
+                    "availableContextTokens": 65536,
+                    "pricing": {
+                        "input": {"usd": 0.375},
+                        "cache_input": {"usd": bad_rate},
+                        "output": {"usd": 1.5},
+                    },
+                },
+            }
+        ]
+    }
+
+    with _patch_models_endpoint(payload):
+        with patch(
+            "routstr.payment.models.async_fetch_openrouter_models",
+            AsyncMock(return_value=[]),
+        ):
+            models = await GenericUpstreamProvider(base_url="http://x").fetch_models()
+
+    model = _model_by_id(models, "venice-badcache")
+    assert model.enabled is True
+    assert model.pricing.prompt == pytest.approx(0.375 / 1_000_000)
+    assert model.pricing.input_cache_read == 0.0
+
+
 # ---------------------------------------------------------------------------
 # native model_spec validation — a bogus native price is not authoritative
 # ---------------------------------------------------------------------------