All benchmarksIndependent evaluation · Epoch AI
ExploitBench
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 Mythos Preview (Early) Limited access | Anthropic | 73.8% | – |
| 2 | GPT-5.5 (unknown thinking) API access | OpenAI | 47.4% | – |
| 3 | Claude Opus 4.7 (no thinking) API access | Anthropic | 26.5% | – |
| 4 | Gemini 3.1 Pro Preview API access | Google DeepMind | 26.1% | – |
| 5 | Claude Sonnet 4.6 (no thinking) API access | Anthropic | 23.6% | – |
| 6 | Kimi K2.6 Open weights (unrestricted) | Moonshot | 18.4% | – |
| 7 | GLM-5.1 Open weights (unrestricted) | Z.ai (Zhipu AI) | 18.1% | – |
| 8 | Claude Haiku 4.5 (unknown thinking) API access | Anthropic | 13.7% | – |
| 9 | MiniMax-M2.7 Open weights (non-commercial) | MiniMax | 13.3% | – |
