Repolode
bytedance

bytedance/UI-TARS-desktop

Open-source multimodal AI agent stack from ByteDance, with a CLI/Web UI (Agent TARS) and a desktop GUI agent app (UI-TARS Desktop).

Open-source alternative to:Anthropic Computer Use

Stars
39,343
Forks
3,997
Open issues
503
Last commit
12 days ago

What is it?

TARS is a multimodal AI agent stack from ByteDance that ships two projects. Agent TARS is a general-purpose agent with a command-line interface and a Web UI, combining GUI and vision capabilities with multimodal LLMs and MCP tools. It can control browsers through visual grounding, DOM, or a hybrid of both, and runs in headful or headless mode.

UI-TARS Desktop is a native desktop application that acts as a GUI agent for a local computer, driven by the UI-TARS model and Seed-1.5-VL/1.6 series models. It takes natural language instructions, reads screenshots, and performs precise mouse and keyboard actions. It also offers local and remote computer and browser operators, and is available for Windows and macOS.

Both projects target developers and users who want an AI agent to operate software through its visual interface rather than through APIs alone. The repository is written in TypeScript and released under Apache-2.0.

Key features

  • 01Command-line interface and Web UI for running the agent, in headful or headless mode
  • 02Browser control via visual grounding, DOM, or a hybrid strategy
  • 03Built on MCP, with support for mounting external MCP servers
  • 04Desktop app for natural-language control of a local computer using screenshots and mouse/keyboard actions
  • 05Local and remote computer and browser operators

Use cases

  • Automating multi-step browser tasks such as booking travel or hotels
  • Letting an AI agent operate desktop applications through screenshots when no API exists
  • Connecting an agent to real-world tools via MCP servers for tasks like generating charts
  • Building GUI automation agents on top of the UI-TARS SDK

Quick start

  1. 1

    Launch Agent TARS with npx

    npx @agent-tars/cli@latest
  2. 2

    Install Agent TARS CLI globally (requires Node.js 22 or later)

    npm install @agent-tars/cli@latest -g
  3. 3

    Run with your model provider and API key

    agent-tars --provider anthropic --model claude-3-7-sonnet-latest --apiKey your-api-key

Things to know

  • Agent TARS CLI requires Node.js 22 or later.
  • Using the agent requires an API key from a model provider such as Volcengine or Anthropic.

Topics

FAQ

What is UI-TARS-desktop?
Open-source multimodal AI agent stack from ByteDance, with a CLI/Web UI (Agent TARS) and a desktop GUI agent app (UI-TARS Desktop). TARS is a multimodal AI agent stack from ByteDance that ships two projects. Agent TARS is a general-purpose agent with a command-line interface and a Web UI, combining GUI and vision capabilities with multimodal LLMs and MCP tools. It can control browsers through visual grounding, DOM, or a hybrid of both, and runs in headful or headless mode.
Is UI-TARS-desktop free to use? What license does it have?
UI-TARS-desktop is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is UI-TARS-desktop still maintained?
Actively maintained. The last commit to UI-TARS-desktop was on Oct 9, 2026.
What can I use UI-TARS-desktop for?
Automating multi-step browser tasks such as booking travel or hotels Letting an AI agent operate desktop applications through screenshots when no API exists Connecting an agent to real-world tools via MCP servers for tasks like generating charts Building GUI automation agents on top of the UI-TARS SDK
How do I install UI-TARS-desktop?
Launch Agent TARS with npx: npx @agent-tars/cli@latest Install Agent TARS CLI globally (requires Node.js 22 or later): npm install @agent-tars/cli@latest -g Run with your model provider and API key: agent-tars --provider anthropic --model claude-3-7-sonnet-latest --apiKey your-api-key
Where can I get UI-TARS-desktop?
The source code and installation instructions are on GitHub: https://github.com/bytedance/UI-TARS-desktop

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects