[Bug]: Model hub not showing customized costs

[xmcp]

What happened?

The "model hub" page in the UI shows costs from the built-in cost map instead of litellm_params. This leads to the displayed costs do not match the actual costs when calling the API.

In the model hub page:

In config.yaml:

  - model_name: azure/gpt-4o-mini
    litellm_params:
      model: azure/gpt-4o-mini
      input_cost_per_token: 0.00000015
      output_cost_per_token: 0.0000006
      ...

  - model_name: ali/qwen-turbo
    litellm_params:
      model: openai/qwen-turbo
      input_cost_per_token: 0.000000042
      output_cost_per_token: 0.000000084
      ...

Relevant log output

N/A

Are you a ML Ops Team?

Yes

What LiteLLM version are you on ?

v1.61.3-stable

Twitter / LinkedIn details

No response