Gemini 3.5 Flash gains Computer Use for browser and desktop automation

Google has now built Computer Use in Gemini 3.5 Flash. Instead of a consumer-facing feature, it is aimed at developers who can create AI agents that can see a screen, reason about what’s on it, and take action—clicking, scrolling, and typing. This isn’t a separate product or an add-on. It’s baked directly into the model devs are already using.

In practice, it shifts Gemini from simply suggesting the next step to helping software carry it out.

Computer use in Gemini 3.5 Flash

Previously limited to Gemini 2.5 Flash, developers can now build custom agents that operate software on your behalf across browser, mobile, and desktop environments. This means browsing the web, filling forms, testing apps, or navigating desktop tools without any custom integrations. The most practical use cases of computer use are enterprise-grade: continuous software testing and knowledge work across professional applications.

Developers and enterprise users can start using the computer used in 3.5 Flash via the Gemini API and Gemini Enterprise Agent platform.

ALSO READ: Claude Tag wants to become your Slack teammate and it’s not just a chatbot window

Google is using targeted adversarial training for computer use in Gemini 3.5 Flash to mitigate prompt injection risk for agents operating in live environments. Two optional enterprise safeguards let businesses require user confirmation before sensitive actions and automatically halt tasks if an injection attempt is detected.

Source: Google

Gemini 3.5 Flash gains Computer Use for browser and desktop automation

Computer use in Gemini 3.5 Flash

More in Gemini

Comments

Leave a Comment