scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-09-10 22:41:41 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 82

Fix #23836: [Bug]: Health checks use max_completion_tokens=1, causing failures for GPT-5 models

Browse source
This commit is contained in:
hannahmadison 2026-03-31 23:26:11 -04:00
parent 33c3f13443
commit 043bc8e7f8
2 changed files with 3 additions and 3 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
litellm/proxy/health_check.py
View file

@ -284,7 +284,7 @@ def _update_litellm_params_for_health_check(
elif "*" not in (
model_info.get("health_check_model") or litellm_params.get("model") or ""
):
litellm_params["max_tokens"] = 1
litellm_params["max_tokens"] = 16
_health_check_model = model_info.get("health_check_model", None)
if _health_check_model is not None:

4
tests/test_litellm/proxy/test_health_check_max_tokens.py
View file

@ -7,14 +7,14 @@ from unittest.mock import AsyncMock, patch, MagicMock
@pytest.mark.asyncio
async def test_update_litellm_params_max_tokens_default():
"""
Test that max_tokens defaults to 1 for non-wildcard models.
Test that max_tokens defaults to 16 for non-wildcard models.
"""
model_info = {}
litellm_params = {"model": "gpt-4"}
updated_params = _update_litellm_params_for_health_check(model_info, litellm_params)
assert updated_params["max_tokens"] == 1
assert updated_params["max_tokens"] == 16
@pytest.mark.asyncio