Write the following content into the Gemini CLI configuration file. Replace <your SU8 API Key> with your own key. The default model is gpt-5.5, but you can also use a model name copied from Models and channels.

{
  "env": {
    "GEMINI_API_KEY": "<your SU8 API Key>",
    "GOOGLE_GEMINI_BASE_URL": "https://www.su8.codes/v1",
    "GEMINI_MODEL": "gpt-5.5"
  }
}

After configuration, fully restart Gemini CLI, then run a simple request to verify it.