Google has added computer use as a built-in tool in Gemini 3.5 Flash, moving the capability into the main Flash model instead of keeping it as a standalone computer-use model. The update is aimed at developers building agents that can work across browser, mobile, and desktop environments.
Google has made computer use a built-in tool in Gemini 3.5 Flash, expanding the model's role in agentic automation. The feature lets developers build agents that can interact with software interfaces across browser, mobile, and desktop environments.
What happened
Google announced the update on June 24, 2026. The company says computer use was previously available as a standalone Gemini 2.5 computer-use model, but is now integrated natively into Gemini 3.5 Flash.
The company says Gemini already supports function calling and built-in tools such as Search and Maps grounding. With computer use added to Flash, Google says developers can use the model to build agents that see, reason, and take action across different computing environments.
Google's Gemini API documentation describes the Computer Use tool as a way to build agents that use screenshots to understand what is on screen and then generate UI actions such as clicks, scrolls, or keystrokes. The documentation says developers still need to implement the client-side execution environment that receives and performs those actions.
Why it matters
Computer use is a key building block for practical agents. Many business tasks still happen inside web apps, desktop tools, forms, dashboards, and internal systems that do not have clean APIs. A model that can understand screens and propose interface actions gives developers another path for automating work that happens inside normal software.
Google framed the update around long-horizon and enterprise automation tasks, including continuous software testing and knowledge work across professional applications. That positioning matters because agent products are moving from chat interfaces toward systems that can operate tools directly. For readers new to the concept, TechPulse has a plain-English guide on what an AI agent is.
The integration also reduces a product boundary for developers. Instead of treating computer use as a separate model choice, teams can evaluate it as one built-in tool available through a general Flash model. That could make experimentation simpler, but it also makes documentation, permission design, and review steps more important.
The safety details are just as important as the feature itself. Google says it uses targeted adversarial training for computer use and is releasing optional enterprise safeguards. Those safeguards include requiring explicit user confirmation for sensitive or irreversible actions and stopping tasks if an indirect prompt injection is identified.
What to watch next
Developers should watch how quickly computer-use workflows move from demos to production apps. Google says developers and enterprises can start using the feature through the Gemini API and Gemini Enterprise Agent Platform, but production use will depend on reliability, policy controls, and how well teams supervise the model's actions.
Security will also stay central. Google's documentation warns that Computer Use is a preview capability that may contain errors and security vulnerabilities. It recommends close supervision for important tasks and avoiding use for critical decisions, sensitive data, or serious irreversible actions.
The broader question is how much agent automation shifts from custom integrations to screen-level workflows. If computer use becomes reliable enough, teams may automate more work inside existing applications without waiting for every vendor to expose a dedicated API.
Source: Google
Frequently Asked Questions
What did Google add to Gemini 3.5 Flash?
Google added computer use as a built-in tool for Gemini 3.5 Flash.
What can Gemini computer use do?
Google says it can help build agents that see a screen and generate UI actions such as clicks, scrolling, or keyboard inputs.
Is Gemini computer use risk-free?
No. Google's documentation says the preview capability may contain errors and security vulnerabilities and recommends close supervision for important tasks.
The Bottom Line
Gemini 3.5 Flash computer use gives developers another route for building agents that operate software interfaces, but the preview status and safety guidance make human oversight a central part of the rollout.
Continue reading related coverage in News or browse the latest articles.