open-computer-use

Why we included this project

Open Computer Use gives you a Linux desktop in the cloud and lets an LLM actually operate it, clicking, typing, and running shell commands. The desktop runs inside an E2B sandbox, keeping it isolated from your own machine, and the live display streams back to you as the agent works. Because you can pause and prompt it mid-task, it feels less like a demo and more like something you could build on. The provider layer is the real strength: a grounding model like OS-Atlas or ShowUI sits alongside separate vision and action models, so switching from Llama to Gemini to GPT-4o or Claude is a config change rather than a rewrite. That flexibility makes it a practical testbed for figuring out which open-source model stack actually handles real GUI work.