Authentication with Gemini CLI begins when you first run the tool, at which point you’ll be prompted to sign in with your personal Google account. This authentication grants you a free Gemini Code Assist license that provides access to Gemini 2.5 Pro and its 1 million token context window. The authentication process is straightforward and designed to get developers up and running quickly. All you need is an email address to get Gemini practically unlimited access in your terminal, making the barrier to entry extremely low for individual developers.
For users who need more control over their API usage or prefer to use specific models, you can generate an API key from Google AI Studio and set it as an environment variable. The process involves creating an API key and then setting it in your terminal using export GEMINI_API_KEY="YOUR_API_KEY". There’s also support for Vertex AI authentication, which requires setting both the API key and an additional environment variable: export GOOGLE_API_KEY="YOUR_API_KEY" and export GOOGLE_GENAI_USE_VERTEXAI=true. This approach is particularly useful for professional developers who need higher rate limits or want to use usage-based billing.
For enterprise users and those with Google Workspace accounts, additional authentication methods are available and documented in the dedicated authentication guide. Professional developers can also obtain Gemini Code Assist Standard or Enterprise licenses for enhanced capabilities and higher usage limits. The authentication system is designed with security in mind, ensuring that access to the AI models is properly controlled while maintaining ease of use. The free tier provides generous limits with 60 model requests per minute and 1,000 model requests per day, which should be sufficient for most individual developers during the preview period. This authentication approach balances accessibility for individual developers with the security and control requirements of enterprise environments.