All benchmarksIndependent evaluation · Epoch AI
OSWorld 2.0
Each model’s result as Epoch AI’s Benchmarking Hub publishes it, run by Epoch or collected from the source it cites. Facthory has not run them.
Data: Epoch AI, ‘Capabilities & benchmarking’, epoch.ai/benchmarks, licensed CC BY 4.0. Read on 16 Nov 2026.
| # | Model | Organization | Result | Evidence |
|---|---|---|---|---|
| 1 | Claude Opus 5 (max) API access | Anthropic | 31.4% | – |
| 2 | Claude Opus 5 (xhigh) API access | Anthropic | 30.2% | – |
| 3 | Claude Opus 5 (high) API access | Anthropic | 29% | – |
| 4 | GPT-5.6 Sol (max) API access | OpenAI | 27.3% | – |
| 5 | Claude Opus 5 (medium) API access | Anthropic | 25.3% | – |
| 6 | Claude Opus 5 (low) API access | Anthropic | 22.3% | – |
| 7 | Claude Opus 4.8 API access | Anthropic | 20.6% | – |
| 8 | Claude Opus 4.7 (max) API access | Anthropic | 18.2% | – |
| 9 | GPT-5.5 (xhigh) API access | OpenAI | 13% | – |
| 10 | Claude Sonnet 4.6 (medium) API access | Anthropic | 9.3% | – |
| 11 | Claude Sonnet 4.6 (max) API access | Anthropic | 8.3% | – |
| 12 | MiniMax-M3 Open weights (restricted use) | MiniMax | 4.6% | – |
| 13 | Kimi K2.6 Open weights (unrestricted) | Moonshot | 4.6% | – |
| 14 | Qwen3.7 Plus API access | Alibaba | 2.8% | – |
