litellm - 💡(How to fix) Fix [Bug]: v1.83.14 auto_router/complexity_router in litellm_params throws "Unmapped LLM provider" 400 error

Check for existing issues

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

What happened?

When configuring the LiteLLM proxy with model: auto_router/complexity_router inside a model's litellm_params, every request returns a 400 BadRequestError:

litellm.BadRequestError: Unmapped LLM provider for this endpoint. You passed model=complexity_router, custom_llm_provider=auto_router

When a fallback is configured, the fallback also fails because the router exposes the pre-alias model name (e.g. claude-sonnet-4-6) in the fallback lookup instead of the resolved group name (internal-ai), so no fallback entry matches: No fallback model group found for original model_group=claude-sonnet-4-6. Fallbacks=[{'internal-ai': ['internal-haiku']}]. Available Model Group Fallbacks=None

Steps to Reproduce

1. Define two backend models in model_list (e.g. internal-haiku, internal-sonnet).

2. Define a router model that uses auto_router/complexity_router: - model_name: internal-ai litellm_params: model: auto_router/complexity_router complexity_router_config: tiers: SIMPLE: internal-haiku REASONING: internal-sonnet complexity_router_default_model: internal-haiku model_info: mode: chat

3. Add a model_group_alias so client-sent model names map to the router:

router_settings: model_group_alias: claude-sonnet-4-6: internal-ai fallbacks: - internal-ai: ["internal-haiku"] 4. Send a chat completion request with model: claude-sonnet-4-6.

1. Observe the 400 BadRequestError, the complexity_router is never invoked and the fallback also fails because the alias name leaks into the fallback lookup.

Relevant log output

What part of LiteLLM is this about?

Proxy

What LiteLLM version are you on ?

v1.83.14

Twitter / LinkedIn details

No response