Chess Puzzles
Games · 2025-12-08
Chess-reasoning benchmark with 100 novel engine-generated puzzles. Scores measure whether models identify the correct moves or solutions.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-5.5 Pro | 64.0 |
| GPT-5.6 Sol Pro | 64.0 |
| GPT-5.4 Pro | 58.6 |
| Gemini 3.1 Pro Preview | 55.0 |
| GPT-5.6 Sol | 55.0 |
| GPT-5.5 | 54.0 |
| GPT-5.6 Terra | 54.0 |
| Gemini 3.5 Flash | 50.0 |
| GPT-5.2 | 49.0 |
| GPT-5.4 | 44.0 |
| Claude Opus 5 | 42.0 |
| Claude Fable 5 | 41.0 |
| Gemini 3.6 Flash | 40.0 |
| GPT-5.6 Luna | 40.0 |
| Kimi K3 | 39.0 |
| Gemini 3 Flash Preview | 38.0 |
| GPT-5 | 37.0 |
| Grok 4.5 | 36.0 |
| Sonnet 5 | 35.0 |
| Opus 4.8 | 34.0 |
| GPT-5.1 | 32.0 |
| Gemini 3 Pro Preview | 31.0 |
| Opus 4.7 | 30.0 |
| GPT-5.4 Nano | 30.0 |
| Grok 4 | 28.0 |
| o3 | 27.0 |
| Kimi K2.6 | 26.0 |
| O4 Mini | 26.0 |
| Grok 4.3 | 25.0 |
| Grok 4.20 | 24.0 |
| Qwen3.7-Max | 22.0 |
| GLM-5.2 | 21.0 |
| Kimi K2.7 Code | 21.0 |
| DeepSeek-V4-Pro | 20.0 |
| Gemini 2.5 Pro | 20.0 |
| gpt-oss-120b | 20.0 |
| Kimi K2 Thinking | 20.0 |
| Qwen3.5-Flash | 20.0 |
| GLM-5.1 | 18.0 |
| GPT-5.4 Mini | 18.0 |