Google has introduced a new open-source AI agent that integrates Gemini’s research, coding, and content creation skills right into developers’ workstations. According to Google, Gemini CLI offers a “fundamental upgrade to your command line experience,” facilitating developers’ use of natural language prompts for code writing and debugging.
According to Google’s announcement, “it offers lightweight access to Gemini, giving you the most direct path from your prompt to our model.” Although it is quite good at coding, Gemini CLI is capable of much more. From content creation and problem resolution to in-depth research and work management, this adaptable, locally available tool can be used for a variety of tasks.
Google’s Gemini 2.5 Pro reasoning model, which is used by Gemini CLI, presently offers a context window of one million tokens (the quantity of data that an AI model can comprehend). According to Google, the Gemini CLI agent enables developers to create photos and videos using its Veo and Imagen AI tools, is connected with Gemini Code Assist, and has built-in support for Model Context Protocol (MCP) and Google Search.
Developers may begin previewing Gemini CLI today, and it is presently free to use with a Gemini Code Assist license that can be acquired using a personal Google account. According to Google, this gives Gemini CLI customers the “largest allowance” available in the market, with a use cap of 60 model queries per minute and 1,000 requests per day.
Google has not stated if the AI agent would continue to be free until it is fully functional or whether there is anything available for customers who surpass those usage caps. It might have an advantage over existing AI coding solutions like GitHub Copilot, Anthropic’s Claude, or even Microsoft’s comparable AI chatbot within Windows Terminal thanks to that huge allowance.
