Chess Puzzles (Epoch AI): leaderboard

Chess puzzle-solving benchmark evaluated by Epoch AI. Tests tactical pattern recognition, calculation depth, and strategic reasoning across a range of puzzle difficulties.

Metric: Accuracy (%). Source: epoch.ai. Status: years away from saturation. 220 models tracked.

Top models

#ModelScore
1GPT-6 (Max)72
2GPT-5.5 Pro (xHigh)64
3GPT-5.4 Pro (xHigh)58.6
4Gemini 3.1 Pro (Preview)55
5GPT-5.6 Sol (Max)55
6GPT-5.5 (xHigh)54
7GPT-5.6 Terra (Max)54
8Gemini 3.5 Flash (High)50
9Gemini 3.1 Pro (Preview) (High)49
10GPT-5.2 (xHigh)49
11Gemini 3.7 Flash (High)47
12Claude Fable 5.1 (Max)47
13DeepSeek V4 Pro (0813) (Max)47
14GPT-5.4 (xHigh)44
15Gemini 3.5 Flash (Minimal)43

Interactive version: theaggregate.ai/benchmark?slug=chess-puzzles-epoch-ai · How It Works · Data refreshed daily, snapshot 2026-09-05.