UI-TARS Desktop

UI-TARS Desktop is ByteDance's open-source app that lets the UI-TARS model operate a computer and a browser.

UI-TARS Desktop is a desktop application around ByteDance's UI-TARS vision-language agent, which clicks and types in graphical interfaces. It ships local computer and browser operators, plus remote computer and remote browser operators that the June 2025 v0.2.0 notes describe as free and needing no extra setup. The same repository also hosts Agent TARS, a CLI and web UI that add GUI and vision skills to a terminal workflow and can use MCP tools. UI-TARS-1.5 and later UI-TARS-2 papers cover GUI, game and tool-use tasks. Midscene.js is the related project for web-only automation.

UI-TARS Desktop at a glance

Best for: Software engineer, Researcher
Runs on: macOS, Windows, open source
Pricing: Free and open source; remote operators are listed as free in v0.2.0
Released: January 2025
Category: AI agent tool
Website: agent-tars.com

Key features

Native desktop GUI agent based on UI-TARS
Local computer and browser operators
Remote computer and browser operators (v0.2.0)
Agent TARS CLI and web UI in the same repo
MCP tool integration in Agent TARS
Open-source models and papers for UI-TARS-1.5 and 2

Use it for

Have an agent click through a desktop app on this Mac
Control a remote browser without writing selectors
Run Agent TARS from a terminal with MCP tools

Alternatives to UI-TARS Desktop

BrowserOS — Runs entirely locally and lets AI agents use real user logins to complete tasks.
Adyen — Share AI agent resources across coding tools and repositories easily.
Mac Computer Use — Enables local desktop automation through native Mac windows and Accessibility controls.
Browser Use Web UI — It supports custom browser integration and persistent sessions.
LobsterAI — Open-source desktop agent with local data storage and remote control via messaging apps.
barn — Monitors multiple AI coding agents on your Mac and jumps to the exact tab when needed.
ruOS — A private agentic desktop that starts instantly in your web browser.
AionUi — Auto-detects and runs local CLI agents like Claude Code and Codex in parallel.
Dev Browser — Persistent browser tabs keep state across scripts so agents do not restart.
Hermes WebUI — Run one server and access your agent from any device.
Offscreen — Runs agent tasks in fresh VMs so your desktop stays free for your own work.
browser-control — Shell-native design with stable element refs and no LLM or framework requirement.

All UI-TARS Desktop alternatives