Chess Puzzles
Epoch AI / Reasoning · Finding correct moves from text descriptions of chess positions
Across e/accCounts toward ranking
Evidence budget1.8%
Upstream data time10/16 06:52
Last successful sync10/17 08:05
What it measures and how
Generated positions are scored against Stockfish answers. Input is position text, so it is not classified as visual understanding.
How this evidence is used
Counts toward ranking; benchmarks in the same task family share a fixed budget.
Benchmark results
Fixed rules select one representative configuration per public model. Anonymous test models are omitted. Original source ranks are retained, with up to 30 models per benchmark.
| Source rank | Source model name | Raw result | Representative configuration |
|---|---|---|---|
| 2 | gpt-6-astra_maxOpenAI | 74.0% | Max reasoning |
| 3 | gpt-6.1-sol_maxOpenAI | 68.0% | Max reasoning |
| 5 | claude-opus-5-5_maxAnthropic | 67.0% | Max reasoning |
| 6 | gemini-3.7-flash_highGoogle | 66.0% | High reasoning |
| 7 | gpt-5.5_xhighOpenAI | 63.0% | xHigh reasoning |
| 8 | gpt-5.5-pro_xhighOpenAI | 62.0% | xHigh reasoning |
| 9 | gpt-5.2-pro-2025-12-11_xhighOpenAI | 60.0% | xHigh reasoning |
| 10 | gpt-5.4-pro-2026-03-05_xhighOpenAI | 59.0% | xHigh reasoning |
| 10 | gpt-5.6-sol_maxOpenAI | 59.0% | Max reasoning |
| 10 | claude-sonnet-5-5_maxAnthropic | 59.0% | Max reasoning |
| 10 | gemini-4-argon_highGoogle | 59.0% | High reasoning |
| 14 | gemini-3.8-flash_highGoogle | 58.0% | High reasoning |
| 15 | gpt-6-sol_maxOpenAI | 56.0% | Max reasoning |
| 16 | claude-haiku-5-5_maxAnthropic | 55.0% | Max reasoning |
| 17 | claude-opus-4-8_maxAnthropic | 52.0% | Max reasoning |
| 18 | gemini-3.1-pro-preview_highGoogle | 51.0% | High reasoning |
| 19 | claude-fable-5-1_maxAnthropic | 48.0% | Max reasoning |
| 19 | deepseek-v4.1-flash_maxDeepSeek | 48.0% | Max reasoning |
| 19 | gpt-5.2-2025-12-11_xhighOpenAI | 48.0% | xHigh reasoning |
| 19 | gpt-5.4-2026-03-05_xhighOpenAI | 48.0% | xHigh reasoning |
| 19 | gpt-5-pro-2025-10-06_highOpenAI | 48.0% | High reasoning |
| 24 | claude-fable-5_maxAnthropic | 47.0% | Max reasoning |
| 24 | gpt-5.6-terra_maxOpenAI | 47.0% | Max reasoning |
| 26 | deepseek-v4-pro-0813_maxDeepSeek | 46.0% | Max reasoning |
| 26 | qwen3.8-max_xhighAlibaba | 46.0% | xHigh reasoning |
| 28 | claude-sonnet-5_maxAnthropic | 45.0% | Max reasoning |
| 28 | gemini-3.5-flash_highGoogle | 45.0% | High reasoning |
| 28 | gpt-5.4-nano-2026-03-17_xhighOpenAI | 45.0% | xHigh reasoning |
| 33 | claude-opus-5_maxAnthropic | 43.0% | Max reasoning |
| 33 | gpt-5.6-luna_maxOpenAI | 43.0% | Max reasoning |
Limitations and attribution
Counts toward ranking; benchmarks in the same task family share a fixed budget.
Data license: CC BY 4.0 · Data evaluated by Epoch AI itself
Results are published by Epoch AI. Raw scores and the e/acc consensus index use different scales and cannot be added directly.