Google has introduced a new AI-based tool — Gemini CLI, which allows access to Gemini models directly from the terminal. This product is aimed at developers working in the command line, enabling them to use natural queries for writing and explaining code, creating new features, finding bugs, executing commands, and automating tasks.
Gemini CLI operates on the Gemini 2.5 Pro model with a context window of 1 million tokens. The tool integrates with “Gemini Code Assist,” supports connections to external databases via the Model Context Protocol, and allows the use of other Google features, such as video generation with Veo, creating research reports with the Deep Research agent, and obtaining up-to-date information through Google search.
The tool is available for free to all users with a personal Google account and a “Gemini Code Assist” license. During preliminary testing, up to 60 model queries per minute and up to 1,000 queries per day can be made, which is the highest free limit among similar solutions. Professional developers have access to support from Google AI Studio or Vertex AI keys, as well as paid licenses.
Gemini CLI is a fully open-source project with source code on GitHub and an Apache 2.0 license. Google encourages developers to contribute to the tool’s development, propose new features, find and fix bugs, and create their own extensions. The tool supports personal settings and integration with scripts, allowing it to be adapted to individual needs.
