Computer Use Agents

ByteDance

UI-TARS

ByteDance's open-weights vision-language agent for GUI control. Prompt templates cover desktop and mobile; browser automation is a separate sibling project.

An open-weights GUI agent model with a companion desktop application — unusual in a field where the capable models are almost all hosted. Because the weights are available, it can run entirely on your own hardware.

Best for

Air-gapped or privacy-constrained deployments where nothing may leave the machine.

Watch out

The directory previously recorded this as controlling both a desktop and a browser. The repository documents desktop and mobile use, and points at separate projects for the desktop application and for web automation, so browser control is not this repository's job. The README's most recent dated update is September 2025.

Changelog

  1. Every entry re-checked: two products moved, one licence was wrong, one surface was wrong

Also in the directory

Agent S Open agentic framework for desktop control, with an emphasis on experience-based memory between runs. Claude Computer Use Screenshot-and-control toolset exposed through the Anthropic API. Moves the mouse, types, and reads the screen of a full desktop environment you host yourself. Cua Containerised desktop sandboxes for agents, with a focus on running macOS and Linux environments safely.