Google has launched a brand new open-source AI agent that brings Gemini’s coding, content material era, and analysis capabilities immediately into builders’ terminals. Gemini CLI gives a “basic improve to your command line expertise,” in response to Google, making it simpler for builders to write down and debug code utilizing pure language prompts.
“It gives light-weight entry to Gemini, supplying you with essentially the most direct path out of your immediate to our mannequin,” Google said in its announcement. “Whereas it excels at coding, we constructed Gemini CLI to take action rather more. It’s a flexible, native utility you should utilize for a variety of duties, from content material era and drawback fixing to deep analysis and process administration.”
Gemini CLI makes use of Google’s Gemini 2.5 Pro reasoning model, which presently helps a 1 million token context window (the quantity of data that an AI mannequin can perceive). Google says the Gemini CLI agent is built-in with Gemini Code Assist, contains built-in Mannequin Context Protocol (MCP) and Google Search assist, and permits builders to generate photos and video utilizing its Veo and Imagen AI instruments.
Gemini CLI is obtainable for builders to preview beginning as we speak, and it’s presently free to make use of via a free Gemini Code Help license that may be obtained through a private Google account. That gives Gemini CLI customers with a utilization restrict of 60 mannequin requests per minute and 1,000 requests per day, which Google says is the “largest allowance” supplied throughout the trade.
Google hasn’t talked about if something is obtainable for customers who exceed these utilization limits, or if the AI agent will stay free-to-use when it turns into absolutely accessible. That beneficiant allowance may give it an edge over different AI coding choices like Anthropic’s Claude Code and the GitHub Copilot Coding Agent, and even Microsoft’s comparable AI chatbot inside Windows Terminal.