Computer use integrated into Gemini 3.5 Flash
The integration of computer use into Gemini 3.5 Flash marks a significant development, enabling the creation of agents capable of operating across various platforms. This functionality, previously exclusive to the standalone Gemini 2.5 computer use model, is now seamlessly incorporated into the core Gemini Flash model.
With this enhancement, Gemini 3.5 Flash offers enhanced functionality for tasks involving agentic computer use. The model excels in executing function calls and utilizing built-in tools such as Search and Maps grounding. This advancement allows developers to create custom agents that can observe, reason, and interact across web browsers, mobile devices, and desktop environments. This capability enhances the performance of long-term and enterprise automation tasks, including continuous software testing and professional application knowledge work.
Both developers and enterprises can now leverage computer use within Gemini 3.5 Flash through the Gemini API and Gemini Enterprise Agent Platform. The 3.5 Flash model utilizes computer use to analyze the Gemini app, providing a categorized list of its features. Additionally, it can audit its documentation for accessibility issues.
To address potential prompt injection risks in live environments, the Gemini 3.5 Flash employs targeted adversarial training. Furthermore, two optional enterprise safeguard systems are available to assist organizations in implementing a robust “defense-in-depth” strategy. This approach encourages developers to integrate these features with secure sandboxing, human oversight, and stringent access controls. More details regarding safety protocols are available in the best practices documentation.
Customers are already benefitting from the capabilities of computer use in the Gemini 3.5 Flash, and feedback indicates positive outcomes. To begin utilizing computer use, ensure your email subscription is confirmed, or subscribe using a different email address.
