🤖 Agentic · llm-stats.com
OSWorld
Real computer-use tasks across operating systems: browsers, office apps and file management.
2
Model diuji
Claude Opus 4.6
Model teratas
72.7%
Skor terbaik
Anthropic
Penyedia
| # | Model | OSWorld |
|---|---|---|
| 1 | Claude Opus 4.6 Anthropic | 72.7% |
| 2 | Claude Sonnet 4.6 Anthropic | 72.5% |
Data benchmark digabungkan dari llm-stats.com. Selengkapnya di metodologi.