Skip to content

LiteLLM - Incorrect baseURL handling in LiteLLM provider when a path is includedΒ #5696

@ChuKhaLi

Description

@ChuKhaLi

App Version

3.23.8 (65a82fd)

API Provider

LiteLLM

Model Used

N/A

Roo Code Task Links (Optional)

No response

πŸ” Steps to Reproduce

  1. Configure the litellm provider in VSCode settings with a baseURL that includes a path, for example http://localhost:4000/litellm
  2. Click Refresh Models
  3. Roo Code UI show error like Failed to fetch LiteLLM models: Failed to fetch LiteLLM models: Unexpected response format.

πŸ’₯ Outcome Summary

Expected: Request should go to http://localhost:4000/litellm/v1/model/info for model fetching and
Actual: Path component is dropped, requests go to http://localhost:4000/v1/model/info

πŸ“„ Relevant Logs or Errors (Optional)

Failed to fetch LiteLLM models: Failed to fetch LiteLLM models: Unexpected response format.

Metadata

Metadata

Assignees

No one assigned

    Labels

    Issue - In ProgressSomeone is actively working on this. Should link to a PR soon.bugSomething isn't working

    Type

    No type

    Projects

    Status

    Done

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions