Google Unveils Gemini 2.5 Computer Use: The Next-Gen AI Model That Takes Action on Web Interfaces

Google has unveiled a preview of its next-generation AI model, Gemini 2.5 Computer Use, a system that can interact with web interfaces as well as understand text and images. Through actions such as clicking, scrolling, typing and dragging, the AI can complete tasks directly on web pages without relying on API integrations, including filling out forms, submitting data, or searching for information.

According to Google, the Gemini 2.5 Computer Use model possesses advanced visual comprehension and reasoning capabilities, enabling it to observe web content and execute user commands with precision. Currently, the model supports 13 core commands and is fully integrated into the Gemini platform, with access for developers via Google AI Studio and Vertex AI.

Gemini 2.5 builds on Project Mariner, which demonstrated autonomous browser task execution, and is described as constrained to browser-level interactions to prioritise security and controllability. The company claims it outperforms other mainstream alternatives in real-world testing and notes the new version can be experimented with and deployed through its platforms.