Google is launching a new agent AI tool that brings Gemini AI models closer to where developers are already coding.
On Wednesday, the company announced the launch of the Gemini CLI, an agent AI tool designed to run locally from the terminal. This new tool can connect Google’s Gemini AI model to a local codebase and allow developers to make natural language requests, such as asking the Gemini CLI to explain the confusion section of the code, create new features, debug code, and run commands.
The Gemini CLI is part of Google’s efforts to help developers use AI models in their coding workflows. Google offers an array of AI coding tools such as Gemini Code Assist and its asynchronous AI coding assistant Jules. However, the Gemini CLI competes directly with other command line AI tools, such as Openai’s Codex CLI and Anthropic’s Claude code. This is a tool that tends to integrate and be faster and more efficient than other AI coding tools.
Since Google launched the Gemini 2.5 Pro in April, its AI models have become a favorite among developers. The popularity of Gemini 2.5 Pro has encouraged the use of third-party AI coding tools such as Cursor and Github Copilot, which have become large-scale businesses. In response, Google has been trying to build a direct relationship with these developers in recent months by offering in-house products.
Most people use Gemini CLI to use it for coding, but the company says it has designed a tool that handles other tasks as well. Developers can tap the Gemini CLI to create videos using Google’s VEO 3 model, generate research reports with the company’s deep search agents through Google search, and access real-time information. Google also says that the Gemini CLI will connect to the MCP server, allowing developers to connect to external databases.
To promote recruitment, Google also has an open-sourcing Gemini CLI under the Apache 2.0 license. This is usually considered one of the most tolerant. The company says it hopes the developers’ network will contribute to the GitHub project.
Google also offers generous use restrictions to promote adoption of the Gemini CLI. Free users can make 60 model requests per minute and 1,000 requests per day. This is almost twice the average number of requests made by developers when using the tool.
The popularity of AI coding tools is rapidly increasing, but there are risks to use them. According to a 2024 survey from Stack Overflow, only 43% of developers trust the accuracy of AI tools. Several studies have shown that code generation AI models may introduce errors or fail to fix security vulnerabilities.
