Google gives Gemini 3.5 Flash computer-use abilities

Google introduced computer-use abilities for Gemini 3.5 Flash, giving developers a model that can operate software interfaces and carry out multi-step tasks across screens. The release is part of the broader shift from chatbots toward agents that can take actions in existing tools. Google framed the feature with controls and safeguards, reflecting the tension between useful automation and the risk of agents making mistakes in live environments.

Key details: Published June 26, 2026, The feature targets Gemini 3.5 Flash, It lets AI agents interact with computer interfaces, Google positioned the release around developer controls and safer task automation.

Why it matters: Computer-use models are the practical bridge from AI answering questions to AI operating real software.