Google introduced its new artificial intelligence (AI) command-line interface tool, dubbed Gemini CLI, on Wednesday. The tool acts as an AI agent and can not only assist developers with their day-to-day coding tasks but can also autonomously complete actions. Since the tool is designed to work directly in the terminal, developers do not have to open a separate window or app to access this feature. Gemini CLI is currently available in preview and can be accessed by anyone for free. The open-source tool is currently being hosted on GitHub.
Google Releases Gemini CLI to Compete With OpenAI
In a blog post, the Mountain View-based tech giant announced the release of Gemini CLI in preview. The command line tool comes two months after OpenAI launched Codex CLI, a similar tool. Google’s offering, however, also comes integrated with Gemini Code Assist, and access to Gemini CLI is entirely free. Those interested in trying out the open-source tool can do so from the GitHub listing.
There are some caveats. While Gemini CLI itself is free, users will only get to use Gemini Code Assist features based on their subscription plan. The tool also has its rate limits. Currently, Google is offering 60 model requests per minute and 1,000 requests per day. However, this should suffice for most developers unless it is used for large enterprise-grade projects.
Gemini CLI is powered by Gemini 2.5 Pro, featuring a context window of one million tokens. The tool is available with the permissive Apache 2.0 license, which allows for both academic and commercial usages. To use it, people can log in with a personal Google account and get a free Gemini Code Assist license.
Additionally, professional developers who either need to run multiple agents simultaneously or want to use specific AI models can use a Google AI Studio or Vertex AI key to get billed based on their usage.
Apart from offering help in writing code, debugging issues, and organising projects with natural language prompts, Gemini CLI also comes with in-built tools that allow users to ground prompts with Google Search, enable Anthropic’s Model Context Protocol (MCP) to connect with external data hubs, and even generate images and videos via the company’s Veo and Imagen models.
