Google has launched a new agentic AI tool called Gemini CLI to bring Gemini AI’s capabilities directly to developer terminals. Once plugged into the Gemini AI models, developers can make requests for debugging, code generation and running commands in natural language easily.
“It provides lightweight access to Gemini, giving you the most direct path from your prompt to our model. While it excels at coding, we built Gemini CLI to do so much more. It’s a versatile, local utility you can use for a wide range of tasks, from content generation and problem solving to deep research and task management,” a blog posted by Google noted.
Gemini CLI uses the Gemini 2.5 Pro for reasoning and coding but users can also plug into other AI models like Imagen and Veo for generating images and videos.
The tool is also integrated with Gemini Code Assist and supports Model Context Protocol (MCP).
Gemini CLI is available for free on Windows, Mac and Linux. Developers just have to sign up for a free Gemini Code Assist license which gives them 60 model requests per minute and 1,000 per day.
If users want to work with multiple agents at the same time or have custom AI models, they can access Gemini CLI via Vertex AI or AI Studio and will be charged based on their usage.
