Introducing computer use in Gemini 3.5 Flash

Google has announced a major enhancement to its Gemini 3.5 Flash model, integrating "computer use" capabilities directly into the AI. This update, revealed on June 24, 2026, allows Gemini 3.5 Flash to act as a more versatile agent, capable of seeing, reasoning about, and taking actions on screens across browser, mobile, and desktop environments. This functionality was previously offered as a separate Gemini 2.5 computer use model, but its native integration into 3.5 Flash simplifies the development process for creating sophisticated AI agents. The core of this new capability lies in its ability to enable AI to interact with graphical interfaces much like a human user would. Developers can now leverage the Gemini API and the Gemini Enterprise Agent Platform to build custom agents that can click, type, scroll, and navigate applications. This opens up a wide array of possibilities for automation, particularly in enterprise settings. For instance, these agents can be deployed for continuous software testing, where they can navigate applications and verify functionality without constant human intervention. Knowledge workers can also benefit by using these agents to complete multi-step browser tasks, fill out forms, extract data from dashboards, and interact with internal tools more efficiently. The integration into Gemini 3.5 Flash means that developers no longer need to manage a two-model workflow, where a separate computer use model was called upon for interface interactions. Instead, computer use can be activated as one of several tools within Flash, alongside existing capabilities like code execution, search, and function calling. This consolidation streamlines the agent-building process and enhances the model's overall performance for long-horizon tasks and enterprise automation. Google emphasizes the safety architecture accompanying this update, having applied targeted adversarial training to ensure secure and reliable operation in sensitive enterprise environments. This focus on safeguards is crucial as AI agents take on more interactive roles within complex systems. The move signifies Google's commitment to advancing agentic AI, pushing towards a future where AI can autonomously perform a broader range of tasks by directly engaging with digital interfaces.