Back to Benchmarks
🤖
OSWorld
OS World
369 GUI/computer use tasks across real OS environments
🤖 Agentic / Tool Use
Best: 76.3%
7 Models Evaluated
Champion
GLM-5
Zhipu AI
89.7%
Best Score
Model Rankings
All models ranked by score on OSWorld. Scores are relative to the best score of 76.3%.
| # | Model | Provider | Score | vs Best |
|---|---|---|---|---|
| 1 | GLM-5 Elo 1454 | Zhipu AI | 89.7% | Best |
| 2 | DeepSeek V4 Pro (Max) | DeepSeek | 76.2% | -0.1% |
| 3 | GPT-5.5 Elo 1478 | OpenAI | 75% | -1.3% |
| 4 | GPT-5.4 Elo 1463 | OpenAI | 75% | -1.3% |
| 5 | Claude Opus 4.6 Elo 1503 | Anthropic | 72.7% | -3.6% |
| 6 | Claude Sonnet 4.6 Elo 1460 | Anthropic | 72.5% | -3.8% |
| 7 | Qwen 3.5 Elo 1450 | Alibaba | 62.2% | -14.1% |
Score Distribution
GLM-5
89.7%
DeepSeek V4 Pro (Max)
76.2%
GPT-5.5
75%
GPT-5.4
75%
Claude Opus 4.6
72.7%
Claude Sonnet 4.6
72.5%
Qwen 3.5
62.2%
Run This Benchmark
Learn how to evaluate your own models on OSWorld.
OSWorld is a standardized benchmark for evaluating AI model capabilities.
To run this benchmark on your own models, visit the official benchmark page or leaderboard for detailed instructions, code, and evaluation scripts.
Data compiled from official model cards, public leaderboards, and independent evaluations.