Evaluation sources Official evaluation
Mystery Game Puzzles
Epoch AI / Reasoning · Derives correct actions from unfamiliar rules
Across NexGateHub AI News Ranked
Evidence budget1.2%
Upstream data time10/15 00:17
Last successful sync10/18 14:05
What it measures and how
Only mean_score under Epoch's standard protocol is used. The unpublished-question limitation remains explicit.
How this evidence is used
Ranked; related question families share a fixed budget.
Evaluation results
Fixed rules select one representative configuration per public model. Anonymous test models are omitted. Original ranks are retained, with up to 30 models per evaluation.
| Source rank | Source model name | Raw result | Representative configuration |
|---|---|---|---|
| 1 | gpt-6-astra_maxOpenAI | 84.0% | Max reasoning |
| 2 | gpt-6.1-sol_maxopenai | 80.0% | Max reasoning |
| 3 | claude-opus-5-5_maxAnthropic | 71.0% | Max reasoning |
| 4 | claude-sonnet-5-5_maxanthropic | 65.0% | Max reasoning |
| 5 | gemini-4-argon_highgoogle | 64.0% | High reasoning |
| 6 | claude-opus-5_maxAnthropic | 59.0% | Max reasoning |
| 7 | gpt-5.6-sol_maxOpenAI | 58.0% | Max reasoning |
| 7 | claude-fable-5-1_maxAnthropic | 58.0% | Max reasoning |
| 9 | gpt-5.5_xhighOpenAI | 56.0% | xHigh reasoning |
| 9 | gpt-6-sol_maxOpenAI | 56.0% | Max reasoning |
| 11 | claude-fable-5_maxAnthropic | 52.0% | Max reasoning |
| 13 | muse-wonder-1.0_max— | 51.0% | Max reasoning |
| 14 | gemini-3.8-flash_highGoogle | 47.0% | High reasoning |
| 15 | deepseek-v4-pro-0813_maxDeepSeek | 43.0% | Max reasoning |
| 15 | deepseek-v4.1-flash_maxDeepSeek | 43.0% | Max reasoning |
| 17 | qwen3.8-max_xhighAlibaba | 38.0% | xHigh reasoning |
| 18 | gemini-3.7-flash_highGoogle | 37.0% | High reasoning |
| 18 | gpt-5.4-2026-03-05_xhighOpenAI | 37.0% | xHigh reasoning |
| 21 | claude-opus-4-8_maxAnthropic | 36.0% | Max reasoning |
| 22 | claude-sonnet-5_maxAnthropic | 35.0% | Max reasoning |
| 22 | gpt-5.6-terra_maxOpenAI | 35.0% | Max reasoning |
| 24 | grok-4.6_xhighxAI | 34.0% | xHigh reasoning |
| 24 | deepseek-v4-flash-0731_maxDeepSeek | 34.0% | Max reasoning |
| 24 | gemini-3.1-pro-preview_highGoogle | 34.0% | High reasoning |
| 27 | glm-5.3_maxZ.ai | 33.0% | Max reasoning |
| 29 | qwen3.7-maxAlibaba | 32.0% | Source default |
| 29 | gemini-3.5-flash_highGoogle | 32.0% | High reasoning |
| 33 | gemini-3.6-flash_highGoogle | 30.0% | High reasoning |
| 33 | claude-haiku-5-5_max— | 30.0% | Max reasoning |
| 35 | o3-2025-04-16_highOpenAI | 29.0% | High reasoning |
Evaluation limits & attribution
Ranked; related question families share a fixed budget.
Data license: CC BY 4.0 · Epoch AI's own evaluation data.
Results are published by Epoch AI. Original scores and the NexGateHub consensus index use different scales and cannot be added directly.