A benchmarking platform for evaluating multimodal AI agents in realistic, open-ended computer environments. It provides a Docker-based environment for testing agent performance across diverse operating systems.
Tech Stack
PythonAI, ML & Data
Why OSWorld?
• Realistic environments
• Standardized metrics
• Multimodal support
Limitations
• Resource intensive
• Complex setup
• Niche research focus
3/5/2026
Last Update
403
Forks
142
Issues
Apache-2.0
License
Financial Leak Detected
Stop the "SaaS Tax"
Your team could be burning cash. Switching to OSWorld instantly boosts your runway.
