Merge pull request #41736 from BerriAI/litellm_mistral_cache_read_pricing

fix(model_prices): add cache-read pricing to Mistral chat models missing it

litellm/model_prices_and_context_window_backup.json

@@ -36785,6 +36785,7 @@
         "supports_tool_choice": true
     },
     "mistral/codestral-mamba-latest": {
+        "cache_read_input_token_cost": 2.5e-08,
         "input_cost_per_token": 2.5e-07,
         "litellm_provider": "mistral",
         "max_input_tokens": 256000,
@@ -36842,6 +36843,7 @@
         "supports_tool_choice": true
     },
     "mistral/devstral-small-latest": {
+        "cache_read_input_token_cost": 1e-08,
         "input_cost_per_token": 1e-07,
         "litellm_provider": "mistral",
         "max_input_tokens": 256000,
@@ -36871,6 +36873,7 @@
         "supports_tool_choice": true
     },
     "mistral/devstral-latest": {
+        "cache_read_input_token_cost": 4e-08,
         "input_cost_per_token": 4e-07,
         "litellm_provider": "mistral",
         "max_input_tokens": 256000,
@@ -36885,6 +36888,7 @@
         "supports_tool_choice": true
     },
     "mistral/devstral-medium-latest": {
+        "cache_read_input_token_cost": 4e-08,
         "input_cost_per_token": 4e-07,
         "litellm_provider": "mistral",
         "max_input_tokens": 256000,
@@ -36986,6 +36990,7 @@
         "source": "https://docs.mistral.ai/models/mistral-embed-23-12"
     },
     "mistral/mistral-medium-3": {
+        "cache_read_input_token_cost": 1.5e-07,
         "input_cost_per_token": 1.5e-06,
         "litellm_provider": "mistral",
         "max_input_tokens": 262144,
@@ -37034,6 +37039,7 @@
         "supports_audio_output": true
     },
     "mistral/voxtral-small-2507": {
+        "cache_read_input_token_cost": 1e-08,
         "input_cost_per_second": 6.666666666666667e-05,
         "input_cost_per_token": 1e-07,
         "litellm_provider": "mistral",
@@ -37049,6 +37055,7 @@
         "supports_tool_choice": true
     },
     "mistral/voxtral-small-latest": {
+        "cache_read_input_token_cost": 1e-08,
         "input_cost_per_second": 6.666666666666667e-05,
         "input_cost_per_token": 1e-07,
         "litellm_provider": "mistral",
@@ -37554,6 +37561,7 @@
         "supports_vision": true
     },
     "mistral/mistral-small": {
+        "cache_read_input_token_cost": 1e-08,
         "input_cost_per_token": 1e-07,
         "litellm_provider": "mistral",
         "max_input_tokens": 32000,
@@ -37680,6 +37688,7 @@
         "supports_vision": true
     },
     "mistral/mistral-tiny": {
+        "cache_read_input_token_cost": 2.5e-08,
         "input_cost_per_token": 2.5e-07,
         "litellm_provider": "mistral",
         "max_input_tokens": 32000,
@@ -37718,6 +37727,7 @@
         "supports_tool_choice": true
     },
     "mistral/open-mistral-nemo": {
+        "cache_read_input_token_cost": 3e-08,
         "input_cost_per_token": 3e-07,
         "litellm_provider": "mistral",
         "max_input_tokens": 128000,
@@ -37803,6 +37813,7 @@
         "supports_vision": true
     },
     "mistral/pixtral-large-latest": {
+        "cache_read_input_token_cost": 2e-07,
         "input_cost_per_token": 2e-06,
         "litellm_provider": "mistral",
         "max_input_tokens": 128000,
@@ -62790,6 +62801,7 @@
         "supports_tool_choice": true
     },
     "mistral/mistral-code-agent-latest": {
+        "cache_read_input_token_cost": 4e-08,
         "input_cost_per_token": 4e-07,
         "litellm_provider": "mistral",
         "max_input_tokens": 256000,

model_prices_and_context_window.json

@@ -36785,6 +36785,7 @@
         "supports_tool_choice": true
     },
     "mistral/codestral-mamba-latest": {
+        "cache_read_input_token_cost": 2.5e-08,
         "input_cost_per_token": 2.5e-07,
         "litellm_provider": "mistral",
         "max_input_tokens": 256000,
@@ -36842,6 +36843,7 @@
         "supports_tool_choice": true
     },
     "mistral/devstral-small-latest": {
+        "cache_read_input_token_cost": 1e-08,
         "input_cost_per_token": 1e-07,
         "litellm_provider": "mistral",
         "max_input_tokens": 256000,
@@ -36871,6 +36873,7 @@
         "supports_tool_choice": true
     },
     "mistral/devstral-latest": {
+        "cache_read_input_token_cost": 4e-08,
         "input_cost_per_token": 4e-07,
         "litellm_provider": "mistral",
         "max_input_tokens": 256000,
@@ -36885,6 +36888,7 @@
         "supports_tool_choice": true
     },
     "mistral/devstral-medium-latest": {
+        "cache_read_input_token_cost": 4e-08,
         "input_cost_per_token": 4e-07,
         "litellm_provider": "mistral",
         "max_input_tokens": 256000,
@@ -36986,6 +36990,7 @@
         "source": "https://docs.mistral.ai/models/mistral-embed-23-12"
     },
     "mistral/mistral-medium-3": {
+        "cache_read_input_token_cost": 1.5e-07,
         "input_cost_per_token": 1.5e-06,
         "litellm_provider": "mistral",
         "max_input_tokens": 262144,
@@ -37034,6 +37039,7 @@
         "supports_audio_output": true
     },
     "mistral/voxtral-small-2507": {
+        "cache_read_input_token_cost": 1e-08,
         "input_cost_per_second": 6.666666666666667e-05,
         "input_cost_per_token": 1e-07,
         "litellm_provider": "mistral",
@@ -37049,6 +37055,7 @@
         "supports_tool_choice": true
     },
     "mistral/voxtral-small-latest": {
+        "cache_read_input_token_cost": 1e-08,
         "input_cost_per_second": 6.666666666666667e-05,
         "input_cost_per_token": 1e-07,
         "litellm_provider": "mistral",
@@ -37554,6 +37561,7 @@
         "supports_vision": true
     },
     "mistral/mistral-small": {
+        "cache_read_input_token_cost": 1e-08,
         "input_cost_per_token": 1e-07,
         "litellm_provider": "mistral",
         "max_input_tokens": 32000,
@@ -37680,6 +37688,7 @@
         "supports_vision": true
     },
     "mistral/mistral-tiny": {
+        "cache_read_input_token_cost": 2.5e-08,
         "input_cost_per_token": 2.5e-07,
         "litellm_provider": "mistral",
         "max_input_tokens": 32000,
@@ -37718,6 +37727,7 @@
         "supports_tool_choice": true
     },
     "mistral/open-mistral-nemo": {
+        "cache_read_input_token_cost": 3e-08,
         "input_cost_per_token": 3e-07,
         "litellm_provider": "mistral",
         "max_input_tokens": 128000,
@@ -37803,6 +37813,7 @@
         "supports_vision": true
     },
     "mistral/pixtral-large-latest": {
+        "cache_read_input_token_cost": 2e-07,
         "input_cost_per_token": 2e-06,
         "litellm_provider": "mistral",
         "max_input_tokens": 128000,
@@ -62790,6 +62801,7 @@
         "supports_tool_choice": true
     },
     "mistral/mistral-code-agent-latest": {
+        "cache_read_input_token_cost": 4e-08,
         "input_cost_per_token": 4e-07,
         "litellm_provider": "mistral",
         "max_input_tokens": 256000,

tests/test_litellm/test_model_prices_schema.py

@@ -3,6 +3,7 @@ from __future__ import annotations
 import importlib.util
 import json
 import re
+from collections.abc import Mapping
 from pathlib import Path
 from types import MappingProxyType
 from typing import Final
@@ -274,3 +275,42 @@ def test_every_bedrock_openai_gpt_row_advertises_xhigh(prices: dict):
         and "xhigh" not in (resolve_supported_reasoning_efforts(entry, deployment_is_mapped=True) or ())
     ]
     assert missing == []
+
+
+def is_active_priced_mistral_chat_row(name: str, entry: Mapping[str, object]) -> bool:
+    input_cost: Final = entry.get("input_cost_per_token")
+    return (
+        name.startswith("mistral/")
+        and entry.get("mode") == "chat"
+        and entry.get("deprecation_date") is None
+        and isinstance(input_cost, (int, float))
+        and input_cost > 0
+    )
+
+
+def cache_read_is_tenth_of_input(entry: Mapping[str, object]) -> bool:
+    cache_read: Final = entry.get("cache_read_input_token_cost")
+    input_cost: Final = entry.get("input_cost_per_token")
+    return (
+        isinstance(cache_read, float)
+        and isinstance(input_cost, (int, float))
+        and 0 < cache_read < input_cost
+        and cache_read == pytest.approx(input_cost / 10)
+    )
+
+
+@pytest.mark.parametrize("path", (PRICES_PATH, BACKUP_PRICES_PATH), ids=("main", "backup"))
+def test_active_mistral_chat_rows_price_cache_reads_below_input(path: Path):
+    """A Mistral chat row without a cache-read rate bills cached prompt tokens at zero, so every
+    active priced row must carry one, and it must be cheaper than a fresh input token. Mistral
+    bills cached tokens at 10% of the input price for every model (docs.mistral.ai/studio/
+    conversations/advanced/prompt-caching, read 2026-09-18), so the ratio is checked as well."""
+    rows: Mapping[str, object] = json.loads(path.read_text())
+    drifted: Final = [
+        f"{name}: cache_read={entry.get('cache_read_input_token_cost')} input={entry.get('input_cost_per_token')}"
+        for name, entry in rows.items()
+        if isinstance(entry, dict)
+        and is_active_priced_mistral_chat_row(name, entry)
+        and not cache_read_is_tenth_of_input(entry)
+    ]
+    assert drifted == []