Google launches Gemini CLI: Open source AI agent transforms terminal development for all programmers free of charge
A new dimension in development tools
Google releases Gemini CLI, an open-source AI agent that works directly in the terminal and gives developers free access to advanced AI features. The tool uses Gemini 2.5 Pro with a 1 million token context window and enables natural language interactions for programming, debugging and task automation.
The application works as a reasoning agent with a ReAct loop (Reason and Act) and executes multi-step workflows. Developers can read files, edit code and execute shell commands using natural language. Multimodal processing through integration with Imagen and Veo models enables the generation of text, images and videos directly in the terminal.
Free availability with generous usage limits of 60 requests per minute and 1,000 requests per day makes the tool freely accessible to individual developers. Installation is done via npm(npm install -g @google/gemini-cli) and requires Node.js version 18 or higher.
Technical architecture and extensibility
The Model Context Protocol (MCP) enables the integration of user-defined tools and data sources. Developers can create their own MCP servers and extend the system for specific use cases. Integrated Google search grounds Prompts with current web data and external information.
Built-in features include codebase analysis through the large context window, automated rebase operations, and text-to-image web application creation. The Apache 2.0 license ensures full transparency and enables code inspection and community modification.
Competitive position and market classification
Google positions Gemini CLI against Anthropics Claude Code and OpenAI’s Codex CLI through free access and open source transparency. 76% of developers already use or plan to use AI coding assistants, according to the Stack Overflow Survey. Terminal-native design addresses productivity hurdles by reducing context switching between development environments.
Integration with VS Code via Gemini Code Assist creates a unified developer experience. Enterprises can scale via Code Assist licenses or Vertex AI billing, while the Apache license provides confidence in proprietary environments.
Executive Summary
-
- Google releases Gemini CLI as a free open source AI agent for terminal development
- 1 million token context window enables full codebase analysis with Gemini 2.5 Pro
- Multimodal functions for text, image and video generation available directly in the terminal
- Free use with 60 requests/minute and 1,000 daily requests for individual developers
- Model Context Protocol allows extension with user-defined tools and data sources
- Apache 2.0 license ensures full code transparency and community contributions
- Integration with Google Search for up-to-date web data and external information sources
- Competitive position against Claude Code and Codex CLI through free open source approach
- Enterprise scaling possible via Code Assist licenses and Vertex AI integration
- Development start via npm installation with Node.js 18 and Google account authentication
Source: Google Blog
