Google distributes Computer Use Gemini 2.5

Gemini 2.5 Computer Use, a new API-accessible paradigm that can operate web browsers and accomplish activities through direct user interface interactions including button clicks and form completion, was launched by Google in preview. The specifics: In order to perform clicks, typing, and navigation commands on its own, the model analyzes screenshots of websites. Computer Gemini 2.5 Utilize beat competitors, such as Claude Sonnet 4.5/4 and OpenAI Computer Using Agent, on both web and mobile benchmarks. Additionally, Google disclosed that variants of the model power Project Mariner and AI Mode tools, demonstrating the highest quality at the lowest latency of the group. Although mainstream users are still in the early stages of fully agentic computer use, the possibilities are developing quickly. There are innumerable time-consuming online actions and workflows that are just begging to be properly automated, beyond the typical examples like scheduling appointments or purchasing.