UI-TARS Desktop
Open computer-use agent that sees and controls your screen.
💸 No earnings reported yet
What it is
ByteDance's open-source (Apache-2.0) computer-use stack, driven by the UI-TARS vision-language model, that reads the screen and clicks, types, and chains multi-step tasks across apps with no API. It composes screen control with structured MCP tool calls and runs as a desktop app or the embeddable Agent TARS framework.
How AI plugs in
The self-hostable answer to Claude computer-use — automate any desktop or web app at compute cost when no API exists, mixing GUI actions with MCP tools.
Alternatives & related tools
Stagehand
Open framework for AI browser automation (by Browserbase).
Suna
Open-source generalist AI agent (Kortix).
Steel
Open-source browser infrastructure for AI agents
Browser Use
Let AI agents control your browser
Anthropic Computer Use
Claude controls a computer — clicks, types, navigates.
Gemini Agent Mode
Google's agent that acts for you — including 24/7 on a Cloud VM (Spark).
★ Reviews
No reviews yet — be the first.Your rating