Scale computer-use 2.0 with open-source drivers, cross-OS sandboxes, and benchmarks for training, evaluation, and data generation.

Agent S: the state-of-the-art open source computer use agents

Bytebot is a self-hosted AI desktop agent that automates computer tasks through natural language commands, operating within a containerized Linux desktop environment.

MagenticLite is an experimental agent that works across the browser and local file system

Fara1.5 – A family of frontier computer use agent models

This is the official website for TuriX Computer-use-Agent

👾 Open Computer Use – Open-Source Alternative to Codex Computer Use

The long-horizon computer-use harness. Run AI agents across desktop apps and the CLI for extended periods while preserving task state and making reliable progress on complex workflows. Features fresh-context execution, durable verified state, independent auditing, recoverable progress, and native Claude Code / Codex / OpenClaw integration.

The first open-source Artificial Narrow Intelligence generalist agentic framework Computer-Using-Agent that fully operates graphical-user-interfaces (GUIs) by using only natural language. Uses Visualization-of-Thought and Chain-of-Thought reasoning to elicit spatial reasoning and perception, emulates, plans and simulates synthetic HID interactions.

UI-Venus is a general-purpose foundation GUI agent for mobile apps, web platforms, and desktop operating systems using only screenshots as input.

Surf is a computer use AI agent powered by OpenAI that interacts with a E2B's virtual desktop environment through natural language instructions

One Computer Use agent for everything (macOS, Windows, Linux, Browser, iOS & Code).

Enable AI to control your desktop, mobile and HMI devices

Accessibility-first browser & computer use for AI agents — one MCP server for every harness (Codex, Claude Code, OpenCode, Antigravity). Instruments the DOM/accessibility tree via CDP instead of screenshot loops: diffed snapshots, stable element refs, trusted input, typed errors. Zero dependencies

OSWorld 2.0: Benchmarking Computer Use Agents on Long-Horizon Real-World Tasks

Stop AI agents from doing things you didn't ask for.

System 1 decision models (Jev, Laya, Cua-S1) as brain for agents: Browser use, computer use, games and robotics

[NeurIPS 2025] Official repository of RiOSWorld: Benchmarking the Risk of Multimodal Computer-Use Agents

[NeurIPS'25 D&B] Mind2Web-2 Benchmark: Evaluating Agentic Search with Agent-as-a-Judge

[ECCV 2026] Official Repository for "DualTAP: A Dual-Task Adversarial Protector for Mobile MLLM Agents"