How to properly use Gemini with ChatDev

[ReppiksProductions]

I can't get ChatDev to work with Gemini. I don't know if I'm calling the wrong models or what. I am running ChatDev on a windows PC via Docker. I edited my .env file as follows:

BASE_URL="https://generativelanguage.googleapis.com/v1beta/openai"
API_KEY="AIza..."

I'm getting 404 error when a node tries to call a model. I've tried gemini-1.5-flash, gemini-1.5-pro, and gemini-2.5-pro. All of which give the following error:
chatdev Model call attempt 1 failed: 404 Not Found. {'message': '', 'status': 'Not Found'}

Can someone please walk me through step by step to get this going?

[zxrys]

We sincerely apologize for the error in our .env.example file regarding the Gemini BASE_URL example. This has now been corrected in the PR #552 . You can actually leave BASE_URL empty, i.e., BASE_URL=, and the system will automatically request the official Gemini endpoint. Otherwise, you can set BASE_URL=https://generativelanguage.googleapis.com.