Data as of Sep 29, 2026 · Based on 2,989 AI responses · See how Parse measures this
OSWorld is a scalable, real computer environment for benchmarking multimodal agents on open-ended tasks across operating systems like Ubuntu, Windows, and macOS. It provides 369 real-world computer tasks with reproducible setup and evaluation scripts, revealing that current AI agents achieve only 12.24% success compared to 72.36% for humans.
Hosted on GitHub Pages
0%No change
of AI answers about OSWorld and its rivals. Week of Sep 21
“OSWorld — University of Hong Kong / Tianbao Xie, Tao Yu and collaborators. Especially relevant if your multimodal agent operates a computer/GUI.”
AI mentioned OSWorld in 0% of answers about OSWorld and its rivals in the week of Sep 21.
