Skip to content
Terminal vibes

Gemini CLI is a free, open source coding agent that brings AI to your terminal

Now, devs who prefer the terminal can get AI assistance, too.

Ryan Whitwam – | 73
Credit: Google
Credit: Google
Story text

Some developers prefer to live in the command line interface (CLI), eschewing the flashy graphics and file management features of IDEs. Google's latest AI tool is for those terminal lovers. It's called Gemini CLI, and it shares a lot with Gemini Code Assist, but it works in your terminal environment instead of integrating with an IDE. And perhaps best of all, it's free and open source.

Gemini CLI plugs into Gemini 2.5 Pro, Google's most advanced model for coding and simulated reasoning. It can create and modify code for you right inside the terminal, but you can also call on other Google models to generate images or videos without leaving the security of your terminal cocoon. It's essentially vibe coding from the command line.

This tool is fully open source, so developers can inspect the code and help to improve it. The openness extends to how you configure the AI agent. It supports Model Context Protocol (MCP) and bundled extensions, allowing you to customize your terminal as you see fit. You can even include your own system prompts—Gemini CLI relies on GEMINI.md files, which you can use to tweak the model for different tasks or teams.

Now that Gemini 2.5 Pro is generally available, Gemini Code Assist has been upgraded to use the same technology as Gemini CLI. Code Assist integrates with IDEs like VS Code for those times when you need a more feature-rich environment. The new agent mode in Code Assist allows you to give the AI more general instructions, like "Add support for dark mode to my application" or "Build my project and fix any errors."

Google wants as many people as possible to use Gemini in their development work, so Gemini CLI is available on Windows, Mac, and Linux—it's also free for almost everyone. Individual developers can sign up to get a free Gemini Code Assist license, which also covers Gemini CLI. The free plan includes 60 model requests per minute and 1,000 per day. Google says this is about twice what its internal team has been using, so almost everyone should be able to integrate Gemini CLI without bumping up against the limit. For professionals who need to run multiple agents at the same time or prefer to use custom models, Gemini CLI supports usage-based billing in Vertex AI or AI Studio.

To get started, just download Gemini CLI from GitHub. That's also where you can report issues and make suggestions to improve Google's AI dev tools.

Photo of Ryan Whitwam
Ryan Whitwam Senior Technology Reporter
Ryan Whitwam is a senior technology reporter at Ars Technica, covering the ways Google, AI, and mobile technology continue to change the world. Over his 20-year career, he's written for Android Police, ExtremeTech, Wirecutter, NY Times, and more. He has reviewed more phones than most people will ever own. You can follow him on Bluesky, where you will see photos of his dozens of mechanical keyboards.
73 Comments
Prev story
Next story