The "Computer Use" feature, previously reserved for a specialized model, becomes a native tool of Gemini 3.5 Flash, allowing agents to see, reason and act on computer, mobile or browser.
Google announced the integration of the "Computer Use" functionality directly into Gemini 3.5 Flash, whereas it previously existed only as an autonomous model derived from Gemini 2.5 Pro. This tool allows developers to build agents that can perceive an interface, reason on its content, and then act on it, whether on computer, mobile or in a browser.
According to Google, this integration significantly improves performance on long-term tasks and corporate automations, such as continuous software testing or some repetitive office work, while maintaining the reduced latency that already distinguished the specialized Computer Use model on web and mobile control benchmarks.
Read the full article on the original source
Go to the source →