Google unveils a new generation of its AI models

Google has announced the launch of a new generation of its intelligent models called “Gemini 2.5 Computer Use”, giving its AI assistant Gemini advanced capabilities to browse and interact directly with websites. This innovation marks a major step toward a new era of intelligent online automation.

According to a report by The Verge, the new model allows Gemini to visit web pages, read their content, and interact autonomously with visual elements — without the need for APIs. The tool can now handle complex tasks such as filling out forms, testing user interfaces, and collecting information directly from websites.

In an official blog post, Google showcased Gemini 2.5 through a series of demo videos illustrating how it can be used to analyze websites, perform interactive actions, and retrieve data in real time from online sources.

This announcement comes amid intensifying competition between major AI companies. OpenAI recently introduced its “GPT Apps” feature, while Anthropic launched similar tools as part of its Claude model.

Google stated that Gemini 2.5 outperforms rival models in both performance and execution speed on computers and smartphones. The tool operates through an integrated browser, enabling interaction with websites without requiring a full operating system or external integrations.

The technology is currently in experimental testing and available to developers via Google AI Studio and Vertex AI, with a global rollout expected soon.