Skip to content

Comments

Explicitly disable VertexAI in Gemini Provider and update Gemini model configs#40

Open
lukasmki wants to merge 3 commits intoagamm:mainfrom
lukasmki:main
Open

Explicitly disable VertexAI in Gemini Provider and update Gemini model configs#40
lukasmki wants to merge 3 commits intoagamm:mainfrom
lukasmki:main

Conversation

@lukasmki
Copy link

Gemini can be accessed by two APIs with different capabilities. Only the Gemini Developer API supports inline and local file batch processing, so I've explicitly disabled vertexai to prevent the client from automatically detecting if you can use Vertex.

This fix also updates the Gemini model configs and... de-hallucinates them. Anyways, here are the references I used for the configs.

@lukasmki lukasmki changed the title Explicitly disable VertexAI in Gemini Provider and de-hallucinate Gemini model configs Explicitly disable VertexAI in Gemini Provider and update Gemini model configs Jan 31, 2026
@lukasmki
Copy link
Author

lukasmki commented Feb 2, 2026

After some testing, tokencost doesn't support Gemini 3 model family and raises a KeyError when running batch.run

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant