UI-TARS Desktop
An open-source desktop AI agent for Windows, macOS and browsers, with visual recognition, mouse and keyboard control, and fully local processing.
UI-TARS Desktop is an open-source AI agent that controls computer interfaces through natural language requests. It runs on Windows and macOS and supports browser use, with operators for both local and remote computers. It's for people who want an agent to carry out tasks in existing apps, such as changing a VS Code setting or checking an issue on GitHub.
The agent uses UI-TARS and Seed-1.5-VL/1.6 series vision-language models to interpret screenshots and recognize what's on screen. It can operate the mouse and keyboard, so its capabilities extend to tasks that require interacting with graphical interfaces. The desktop app displays feedback and task status while the agent works.
Fully local processing is a stated feature of the desktop application, making it relevant to users looking for local AI that can act on their computer. Remote computer and browser operators are also available. The project uses the Apache 2.0 license.
UI-TARS Desktop shares the TARS stack with Agent TARS, but the two have different interfaces and capabilities. UI-TARS Desktop provides the native desktop GUI agent. Agent TARS is the companion CLI and web interface, with browser control through visual interaction, the page DOM or a combination of both, plus MCP connections to external tools.
