← Benchmarks
OSWorld-Verified
OSWorld-Verified is a verified subset of OSWorld, a scalable real computer environment for multimodal agents supporting task setup, execution-based evaluation, and interactive learning across Ubuntu, Windows, and macOS.
id osworld-verified · max 1 · 26 models reported
| # | Model | Score |
|---|---|---|
| 1 | Qwen3.8 Max Alibaba Cloud / Qwen Team · open | 0.86 |
| 2 | Claude Fable 5 Anthropic | 0.85 |
| 3 | Qwen3.8-27B Alibaba Cloud / Qwen Team · open | 0.84 |
| 4 | Claude Opus 4.8 Anthropic | 0.83 |
| 5 | Gemini 3.6 Flash Google | 0.83 |
| 6 | MiMo-V2.6-Pro Xiaomi · open | 0.82 |
| 7 | Claude Sonnet 5 Anthropic | 0.81 |
| 8 | MiMo-V2.6-Flash Xiaomi · open | 0.81 |
| 9 | Muse Spark 1.1 Meta | 0.81 |
| 10 | Claude Mythos Preview Anthropic | 0.80 |
| 11 | GPT-5.5 OpenAI | 0.79 |
| 12 | Gemini 3.5 Flash Google | 0.78 |
| 13 | Claude Opus 4.7 Anthropic | 0.78 |
| 14 | GPT-5.4 OpenAI | 0.75 |
| 15 | Gemini 3.5 Flash-Lite Google | 0.74 |
| 16 | Qwen3.7-Plus Alibaba Cloud / Qwen Team | 0.73 |
| 17 | Kimi K2.6 Moonshot AI · open | 0.73 |
| 18 | GPT-5.4 mini OpenAI | 0.72 |
| 19 | MiniMax M3 MiniMax · open | 0.70 |