mondaychen/gemini-cli

An open-source AI agent that brings the power of Gemini directly into your terminal.

Gemini CLI (OpenAI compatible)

This is a fork of the Gemini CLI that allows you to use any LLM API, as long as it's OpenAI compatible.

To see how to use Gemini CLI, see the official Gemini CLI documentation.

Usage

Clone the repo and run the following command:

npm install
npm run build
npm install -g .

In the project root where you run Gemini CLI, create a .env file and set the following variables:

OPENAI_API_KEY=
OPENAI_BASE_URL= # optional, defaults to https://api.openai.com/v1
MAIN_MODEL=gpt-4o # your choice of model
FLASH_MODEL=gpt-4o-mini # your choice of fast/mini model for simple tasks

Then run gemini to start the CLI. You'll choose "OpenAI" when prompted. If you used Gemini CLI before, you'll need to type /auth to reset the auth method.

Limitations

• Tokens are not counted accurately.