UI-TARS

UI-TARS is a multimodal vision-language AI agent designed to interact with graphical user interfaces in a human-like way. It combines visual perception, reasoning, grounding, memory and action within an agent model, allowing it to understand screenshots and perform actions such as clicking, typing, scrolling, dragging and using keyboard shortcuts. It can automate desktop, web and mobile GUI tasks and is suitable for computer-use agents, browser automation, software testing and research. ByteDance also provides UI-TARS-desktop, an application that lets users control a computer using natural-language instructions. UI-TARS can be deployed locally or through compatible cloud/model endpoints. Availability: The project is currently active and open source; there is no evidence of country-specific unavailability, including India.