UI-TARS-desktop

About

UI-TARS-desktop is an open-source native desktop application developed by Bytedance, providing a powerful GUI Agent based on the UI-TARS multimodal model. It enables AI to perceive and interact with graphical user interfaces just like a human, supporting both local and remote computer/browser operations. Built on the Model Context Protocol (MCP), it integrates seamlessly with real-world tools. The app features a redesigned UI and leverages the UI-TARS-1.5 model for high-precision vision-based control and task automation.

Features

• Native GUI Agent Interaction
• Remote Computer & Browser Operation
• Vision-based Multimodal Control
• MCP Protocol Integration
• End-to-end Task Automation

Supported Platforms