Google Releases Gemini CLI: A Terminal-Native AI Agent with Open-Source Access

Google has launched Gemini CLI, a free and open-source command-line interface powered by its Gemini 2.5 Pro model. The tool supports a 1 million-token context window, large enough to ingest entire codebases or deeply nested repositories, enabling developers to debug, refactor, or generate code with full project awareness. Beyond code, Gemini CLI can automate workflows, perform research, and produce media via Veo and Imagen, all from the terminal. The tool is licensed under Apache 2.0, meaning developers can inspect, extend, and embed it into their own pipelines without vendor restrictions. Gemini CLI connects natively with Gemini Code Assist, Google Search, and the Model Context Protocol, offering a cohesive agent experience across tools. In preview mode, users get a generous quota: 60 requests per minute and 1,000 per day. Gemini CLI marks a shift toward agent-style developer tools that go beyond suggestion engines. Positioned alongside OpenAI’s Codex and Anthropic’s Claude Code, it’s designed for deep integration, not just autocomplete—making it a flexible, script-friendly solution for DevOps, AI engineers, and anyone building at the terminal level.