scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-11-14 09:03:16 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 214

fix(mistral): declare low, medium and high reasoning effort for mistral-large-4 (#51693)

Browse source 
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
This commit is contained in:
kenneth-berri 2026-11-13 12:20:18 -08:00 • committed by GitHub
parent 853773e75a
commit 534ef43526
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
3 changed files with 13 additions and 7 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

6
litellm/model_prices_and_context_window_backup.json
View file

@ -44324,7 +44324,8 @@
"mode": "chat",
"output_cost_per_token": 2.1e-06,
"reasoning_effort_levels": [
"none",
"low",
"medium",
"high"
],
"source": "https://docs.mistral.ai/models/mistral-large-4-0",
@ -44345,7 +44346,8 @@
"mode": "chat",
"output_cost_per_token": 2.1e-06,
"reasoning_effort_levels": [
"none",
"low",
"medium",
"high"
],
"source": "https://docs.mistral.ai/models/mistral-large-4-0",

6
model_prices_and_context_window.json
View file

@ -44324,7 +44324,8 @@
"mode": "chat",
"output_cost_per_token": 2.1e-06,
"reasoning_effort_levels": [
"none",
"low",
"medium",
"high"
],
"source": "https://docs.mistral.ai/models/mistral-large-4-0",
@ -44345,7 +44346,8 @@
"mode": "chat",
"output_cost_per_token": 2.1e-06,
"reasoning_effort_levels": [
"none",
"low",
"medium",
"high"
],
"source": "https://docs.mistral.ai/models/mistral-large-4-0",

8
tests/unit/test_mistral_large_4_model_metadata.py
View file

@ -11,6 +11,8 @@ from litellm.llms.mistral.chat.transformation import MistralConfig
REPO_ROOT = Path(__file__).parents[2]
MODELS = ("mistral/mistral-large-4", "mistral/mistral-large-4-0")
# Probed against api.mistral.ai on 2026-11-13: both ids take these three levels and answer 400 to none.
ACCEPTED_EFFORTS: Final = ("low", "medium", "high")
@pytest.fixture(autouse=True)
@ -34,7 +36,7 @@ def test_metadata_and_backup(model):
assert info["mode"] == "chat"
assert info["max_input_tokens"] == 1048576
assert 0 < info["cache_read_input_token_cost"] < info["input_cost_per_token"] < info["output_cost_per_token"]
assert info["reasoning_effort_levels"] == ["none", "high"]
assert info["reasoning_effort_levels"] == list(ACCEPTED_EFFORTS)
for capability in (
"supports_vision",
"supports_reasoning",
@ -49,7 +51,7 @@ def test_metadata_and_backup(model):
@pytest.mark.parametrize("model", MODELS)
@pytest.mark.parametrize("effort", ["none", "high"])
@pytest.mark.parametrize("effort", ACCEPTED_EFFORTS)
def test_reasoning_effort(model, effort):
config = MistralConfig()
assert "reasoning_effort" in config.get_supported_openai_params(model)
@ -92,7 +94,7 @@ def test_large_3_alias_is_not_retargeted():
@pytest.mark.parametrize("model", MODELS)
@pytest.mark.parametrize("sync_mode", [True, False])
@pytest.mark.parametrize("effort", ["none", "high"])
@pytest.mark.parametrize("effort", ACCEPTED_EFFORTS)
@pytest.mark.asyncio
async def test_completion_contract(model, sync_mode, effort, respx_mock):
"""Exercise the real adapter against explicitly mocked Mistral HTTP responses."""