Agent infrastructure
OSWorld
xlang-ai/OSWorld
OSWorld is a benchmark and real desktop environment for evaluating multimodal computer-use agents, with support for VMware, Docker, AWS, and other backends for reproducible task evaluation.
Topics
Related projects
cl-bench592A benchmark series for context learning that evaluates reasoning and learning ability through workplace and everyday tasks with thousands of detailed annotation rules.
OpenHands93kA self-hosted control center that orchestrates multiple coding agents and automations.
ai27.9kVercel's provider-agnostic TypeScript toolkit for building AI apps and agents with unified models, tools, and chat UI.
owl20.2kA general multi-agent workforce that automates real-world tasks using search, browser, and code tools.
DeepResearch20.1kAn open-weight 30B deep-research model release for long-horizon information seeking, with ReAct inference scripts, benchmarks, and demos.
agent-lightning18.9kA lightweight RL framework for training AI agents directly with their real harnesses and environments.