Gemini Model Errors Persist Despite Overridden Base URL for OpenAI API

[edisonzf2020]

1. If you can, please include a screenshot of your problem

2. Please include the name of your operating system macos Apple M2 Ultra

3. If you can, steps to reproduce are super helpful When using the OpenAI API with an overridden base URL, the Gemini model encounters errors due to it being directed to use the OpenAI base URL instead of its own dedicated channel. This misdirection leads to incorrect model selection and potential errors in the system.

Proposed Solution: To address this issue, I suggest modifying the cursor model setup as follows:

1. Separate Override Settings: Ensure each API has its own override base URL setting, allowing for distinct configurations per API.

2. Model Naming Convention: In each API, there are own model names settings.