UI-TARS Desktop
An open-source desktop GUI agent from ByteDance that operates your computer using multimodal AI models.
These buttons open the developer's own site, repository or store listing in a new tab. wares.gg does not host downloads.
About UI-TARS Desktop
UI-TARS Desktop is a desktop application that lets a multimodal AI model see the screen and control the computer to carry out tasks you describe. It is built on ByteDance's UI-TARS models.
The repository has grown into a broader open-source multimodal agent stack that connects AI models with agent infrastructure, with the desktop app kept alongside other packages and examples. It suits people who want to experiment with computer-use agents on their own machine.
Strengths
- Controls the desktop through visual understanding of the screen
- Open source with an active, widely followed repository
- Part of a larger agent toolkit
Limitations
- Needs access to a capable vision model
- Giving an agent control of your computer carries risk
Details
- Pricing
- FreeFree and open source; model hosting or API costs may apply.
- License
- Open source, license not stated
- Developer
- ByteDance
- Platforms
- Windows, macOS
- How it runs
- Downloadable app
- Best suited for
- Experimenting with AI computer-use agents on your own desktop
- Categories
- Automation, AI agents
- Last verified
- Added
Alternatives to UI-TARS Desktop
Compare allSoftware that can replace UI-TARS Desktop for an important use case, and what changes if you switch.
Cua
Open-source infrastructure for running computer-use AI agents in sandboxed virtual machines across operating systems.
Cua runs computer-use agents inside sandboxed virtual machines across several operating systems, keeping agents off your own desktop, but targets developers.
Agent Zero
A self-hosted personal AI agent framework that uses the computer as a tool and runs in Docker.
Agent Zero is a self-hosted Docker agent that uses the computer as a tool through code and a web interface rather than visual screen control.
GAIA
AMD's open-source app for building and running local AI agents on Ryzen AI PCs.
GAIA is AMD's open-source app for running local agents on Ryzen AI PCs, Windows only, oriented toward building agents rather than screen control.
Goose
Run an extensible AI assistant on your desktop or in a terminal.
Goose is an Apache-licensed desktop and terminal assistant that acts through MCP extensions rather than seeing and clicking the screen, and adds Linux support.
Similar software
Related functionality, not necessarily a direct replacement.
Browser Use
Open source framework that lets AI agents control a web browser.
Screenpipe
Records your screen and audio continuously on your own computer so AI agents can search it.
OpenManus
A general-purpose AI agent written in Python, built as an open alternative to Manus.
Skyvern
Open-source AI agents that automate browser workflows such as logins, form filling and data extraction.
Fellou
A desktop web browser with a built-in AI agent that carries out multi-step tasks across websites.
