Gemini Computer Use is generally available

Google's Gemini API changelog for 30 October 2026 says the Computer Use tool is generally available on the Interactions API and the GenerateContent API. The docs describe it as a way to build browser, mobile, and desktop agents. The model looks at screenshots and returns UI actions such as mouse clicks and keyboard input. The caller has to run those actions. Google's docs say that is the same split as function calling: the model proposes the step, and your environment performs it.

The docs list four models. Gemini 4 Argon (gemini-4-argon) is described as the most capable option for complex reasoning and agentic UI work. Gemini 3.8 Flash (gemini-3.8-flash) is the recommended model, for accurate UI interaction and tool calling. Gemini 3.5 Flash-Lite (gemini-3.5-flash-lite) is the low-latency option. Gemini 3 Flash Preview (gemini-3-flash-preview) is still a preview. The docs also say Gemini 4 and Gemini 3.x can attach an intent to each action, set safety policies, and opt in to screenshot scanning for hidden instructions. A browser example enables the tool with environment set to browser.

Who it's for: a developer who wants a Gemini model to propose clicks and keystrokes, and who will run those actions in their own browser or desktop.

Fits: Best for Vibe Coding stays Cursor, for an edit you can reject inside the project. Gemini stays third on Best for Writing. Computer Use is an API tool, not that chat ranking. The Gemini start guide and the Gemini alternatives are unchanged. The verdicts are unchanged.