Gemini 3.5 Flash supports Computer use capabilities

The Computer use feature has been integrated as a default tool in Gemini 3.5 Flash. Previously, a separate Gemini 2.5 model was required, but it is now supported directly within the Flash model, enabling faster and more efficient agent construction.

Developers can use this feature to build custom agents that can see, reason, and act across browser, mobile, and desktop environments. This is optimized for enterprise automation tasks with long-term goals, such as software test automation and knowledge work leveraging specialized applications.

Key features and safety measures:

• Integrated capabilities: Provides powerful agent performance when combined with Search, Maps grounding, and Function calling
• Enhanced security: Applies adversarial training to prevent prompt injection
• Enterprise safety systems: Requires user approval for sensitive tasks and automatically stops operations when indirect prompt injection is detected

It is currently available immediately through the Gemini API and Gemini Enterprise Agent Platform.