litellm - 💡(How to fix) Fix [Bug]: /health returns empty healthy_endpoints and unhealthy_endpoints for hosted_vllm models on v1.84.x

Check for existing issues

• I have searched the existing issues and checked that my issue is not a duplicate.

What happened?

LiteLLM Version

v1.84.x

What happened?

/health returns:

json { "healthy_endpoints": [], "unhealthy_endpoints": [] }

for all hosted_vllm models.

However:

• /v1/models works
• /v1/model/info works
• /v1/chat/completions works

The models are reachable and inference succeeds correctly.

Reproduction Steps

Configuration:

yaml model_list: - model_name: Gpt-oss-120b litellm_params: model: hosted_vllm/Gpt-oss-120b api_base: http://vllm:8000/v1

Health check:

bash curl "http://litellm:4000/health"

or

bash curl "http://litellm:4000/health?model=Gpt-oss-120b"

returns:

json { "healthy_endpoints": [], "unhealthy_endpoints": [] }

Expected Behavior

The endpoint should populate either:

• healthy_endpoints
• or unhealthy_endpoints

Relevant Logs

text GET /health?model=Gpt-oss-120b HTTP/1.1" 200 OK

No provider errors appear in the logs.

Additional Context

• Self-hosted vLLM
• HuggingFace weights
• Inference works correctly through /v1/chat/completions
• The issue started after upgrading from v1.80.0 to v1.84.x

Steps to Reproduce

Relevant log output

What part of LiteLLM is this about?

No response

What LiteLLM version are you on ?

v1.84.0

Twitter / LinkedIn details

No response