Gemini CLI

Gemini CLI is Google's official Gemini command-line tool. TokOne supports the Google protocol (/v1beta), so you can use Gemini models with the config below.

Install

npm install -g @google/gemini-cli

Verify:

gemini --version

Configure

Gemini CLI uses two environment variables: GEMINI_API_KEY and GEMINI_BASE_URL.

Base URL is /v1beta

Google's official protocol root is /v1beta (not /v1), and TokOne follows the same path.

macOS / Linux

export GEMINI_API_KEY="sk-xxx"
export GEMINI_BASE_URL="https://api.tokone.ai/v1beta"

Windows Terminal

$env:GEMINI_API_KEY = "sk-xxx"
$env:GEMINI_BASE_URL = "https://api.tokone.ai/v1beta"

To persist it, add it to ~/.zshrc (macOS) / ~/.bashrc (Linux) / Windows user-level environment variables.

Verify

gemini "Reply with OK"

A reply means the config works.

FAQ

404 even though the key is correct

The Base URL is missing /v1beta. Google's protocol root differs from OpenAI's, so you must include /v1beta explicitly.

model not found

Check the model ID (e.g. gemini-3.5-flash); see the Models page for the available list.

SSL / certificate errors

Make sure the protocol is https://, not http://.

Next steps

• Models — available Gemini models
• API Endpoints — Gemini-compatible protocol notes