Skip to content

Google Unveils Gemini 2.5 Computer Use Model for AI-Powered Assistants and UI Testing

Google's new model is already in production for UI testing. It's optimized for web browsers and shows promise for mobile UI control. Safety features are integrated to mitigate risks.

In this picture we can see a web page, in the web page we can find some text and a machine.
In this picture we can see a web page, in the web page we can find some text and a machine.

Google Unveils Gemini 2.5 Computer Use Model for AI-Powered Assistants and UI Testing

Google has unveiled the Gemini 2.5 Computer Use model, a specialized tool built on the visual understanding and reasoning prowess of Gemini 2.5 Pro. This model is designed to empower personal assistants, automate workflows, and streamline UI testing.

Early testers have put the model through its paces, using it to power personal assistants, automate workflows, and test user interfaces. Results have been promising, with Google teams already employing it in production for UI testing.

The Gemini 2.5 Computer Use model is optimized for web browsers and shows potential for mobile UI control tasks. However, it's not yet primed for desktop OS-level control. Safety features are integrated into the model to mitigate risks, and developers have access to safety controls to prevent high-risk actions. Google's documentation offers additional safety measures and best practices.

The model's core capabilities are accessible via the new tool in the Gemini API. It operates in a loop, analyzing inputs and generating responses that represent UI actions. In tests, it outperformed leading alternatives on web and mobile control benchmarks with lower latency.

The Gemini 2.5 Computer Use model is now in public preview, available via the Gemini API on Google's AI Studio and Vertex AI. Developers can start building their own agent loops with access to demos and documentation. Before its public release, the model was used in closed test phases and preview environments, accessible via the Gemini API, Google AI Studio, mobile/web apps, and Vertex AI for enterprise customers.

Read also:

Latest