Benchmarks / Reasoning

Chess Puzzles (Epoch AI)

Epoch AI's chess puzzles: the model finds the best move in positions with a single correct line. Score is the share of puzzles solved. Data: Epoch AI, CC BY.

83models
–Open Index weight
–last updated
epoch-taskspublisher

Chess Puzzles (Epoch AI)

Higher is better · best effort per model

Chart settings
83 of 83 selected

full table below

#ModelEffortScoreCost / taskHarness
1GPT-6 Astra Promax76.0%–Epoch AI
2GPT-6 Astramax74.0%–Epoch AI
3GPT-5.6 Sol Promax68.0%–Epoch AI
4GPT-6.1 Solmax68.0%–Epoch AI
5Opus 5.5max67.0%–Epoch AI
6Gemini 3.7 Flashhigh66.0%–Epoch AI
7GPT-5.5xhigh63.0%–Epoch AI
8GPT-5.5 Proxhigh62.0%–Epoch AI
9GPT-5.2 Proxhigh60.0%–Epoch AI
10GPT-5.4 Proxhigh59.0%–Epoch AI
11Gemini 4high59.0%–Epoch AI
12GPT-5.6 Solmax59.0%–Epoch AI
13Sonnet 5.5max59.0%–Epoch AI
14Gemini 3.8 Flashhigh58.0%–Epoch AI
15GPT-6 Solmax56.0%–Epoch AI
16Haiku 5.5max55.0%–Epoch AI
17Opus 4.8max52.0%–Epoch AI
18Gemini 3.1 Prohigh51.0%–Epoch AI
19GPT-5.2xhigh48.0%–Epoch AI
20GPT-5 Prohigh48.0%–Epoch AI
21GPT-5.4xhigh48.0%–Epoch AI
22Fable 5.1max48.0%–Epoch AI
23DeepSeek V4.1 Flashmax48.0%–Epoch AI
24GPT-5.6 Terramax47.0%–Epoch AI
25Fable 5max47.0%–Epoch AI
26Qwen 3.8 Maxxhigh46.0%–Epoch AI
27DeepSeek V4 Pro 0813max46.0%–Epoch AI
28GPT-5.4 Nanoxhigh45.0%–Epoch AI
29Gemini 3.5 Flashhigh45.0%–Epoch AI
30Sonnet 5max45.0%–Epoch AI
31GPT-5.6 Lunamax43.0%–Epoch AI
32Opus 5max43.0%–Epoch AI
33GPT-5.1high42.0%–Epoch AI
34Gemini 3.6 Flashlow42.0%–Epoch AI
35Opus 4.7max41.0%–Epoch AI
36Grok 4.7xhigh41.0%–Epoch AI
37Gemini 3 Flashhigh39.0%–Epoch AI
38GLM 5.3max37.0%–Epoch AI
39Kimi K3high37.0%–Epoch AI
40o3medium37.0%–Epoch AI
41Muse Spark 1.3max37.0%–Epoch AI
42GPT-5high36.0%–Epoch AI
43GPT-6 Lunamax36.0%–Epoch AI
44Grok 4.5high35.0%–Epoch AI
45Muse Spark 1.1xhigh34.0%–Epoch AI
46Mistral Large 4high33.0%–Epoch AI
47Grok 4.6xhigh33.0%–Epoch AI
48DeepSeek V4 Flash 0731max32.0%–Epoch AI
49Grok 4.20xhigh29.0%–Epoch AI
50Muse Spark 1.2xhigh27.0%–Epoch AI
51Qwen 3.7 Maxmax26.0%–Epoch AI
52GPT-5 Minihigh24.0%–Epoch AI
53GPT-5.4 Minihigh23.0%–Epoch AI
54Sonnet 4.6max23.0%–Epoch AI
55GLM 5.2max22.0%–Epoch AI
56o4 minimedium22.0%–Epoch AI
57GPT-5 Nanohigh22.0%–Epoch AI
58GLM 4.7none21.0%–Epoch AI
59GLM 5.3 Flashmax21.0%–Epoch AI
60Opus 4.6max21.0%–Epoch AI
61Gemini 3.5 Flash-Litelow21.0%–Epoch AI
62Inklingxhigh21.0%–Epoch AI
63GLM 5.1max20.0%–Epoch AI
64Kimi K2.6max20.0%–Epoch AI
65DeepSeek V4 Promax19.0%–Epoch AI
66Grok 4.3high18.0%–Epoch AI
67o3 minihigh15.0%–Epoch AI
68Kimi K2.7 Codemax14.0%–Epoch AI
69Opus 4.5high14.0%–Epoch AI
70Qwen 3.5 Flashmax14.0%–Epoch AI
71Qwen 3.6 Plusmax14.0%–Epoch AI
72MiniMax M3max13.0%–Epoch AI
73Qwen 3.6 Flashmax12.0%–Epoch AI
74o1high10.0%–Epoch AI
75Sonnet 4.5high10.0%–Epoch AI
76Haiku 4.5xhigh8.0%–Epoch AI
77Qwen 3.5 Plusnone8.0%–Epoch AI
78GPT-4ohigh6.0%–Epoch AI
79GPT-4o minihigh3.0%–Epoch AI
80GPT-4.1high3.0%–Epoch AI
81Opus 3none3.0%–Epoch AI
82GPT-4.1 Minihigh1.0%–Epoch AI
83GPT-4.1 Nanohigh0.0%–Epoch AI