Benchmark Almanac

Chess Puzzles

Claude Fable 5.2 79% leads since November 2026
Not in model cards

Models find the single best next move in 100 chess puzzles that Epoch generated with a chess engine. Scored by the share solved.

Made by
Epoch AI
Released
December 2025
Kind
Puzzles and chess
Level
Competition
Scored
Checked automatically
Answers
Short answer
Tasks
All public
Strongest model tested
Claude Fable 5.2 #1 on the capability index
Last model added
November 2026
Scores refreshed
November 12th, 2026
93
Models scored in Epoch AI’s runs

Claude Fable 5.2 leads with 79%

Best score per model from Epoch AI’s own runs. Where a model ran with several settings or harnesses, we keep its best.

93 models. Scroll sideways for the rest.

  1. 79.0 Claude Fable 5.2
  2. 78.0 Gemini 4 Carbon
  3. 76.0 GPT-6 Astra
  4. 68.0 GPT-5.6 Sol
  5. 68.0 GPT-6.1 Sol
  6. 67.0 Claude Opus 5.5
  7. 66.0 Gemini 3.7 Flash
  8. 63.0 GPT-5.5
  9. 62.0 GPT-5.5 Pro
  10. 60.0 GPT-5.2 Pro
  11. 59.0 GPT-5.4 Pro
  12. 59.0 Claude Sonnet 5.5
  13. 59.0 Gemini 4 Argon
  14. 58.0 Gemini 3.8 Flash
  15. 56.0 GPT-6 Sol
  16. 55.0 Claude Haiku 5.5
  17. 52.0 Claude Opus 4.8
  18. 51.0 Gemini 3.1 Pro
  19. 48.0 GPT-5 Pro
  20. 48.0 GPT-5.2
  21. 48.0 GPT-5.4
  22. 48.0 Claude Fable 5.1
  23. 48.0 DeepSeek V4.1 Flash
  24. 47.0 Claude Fable 5
  25. 47.0 GPT-5.6 Terra
  26. 46.0 Qwen 3.8 Max
  27. 46.0 DeepSeek V4 Pro 0813
  28. 46.0 DeepSeek V4.1 Flash 1017
  29. 45.0 GPT-5.4 Nano
  30. 45.0 Gemini 3.5 Flash
  31. 45.0 Claude Sonnet 5
  32. 45.0 Kimi K3.1 Pro
  33. 43.0 GPT-5.6 Luna
  34. 43.0 Claude Opus 5
  35. 42.0 GPT-5.1
  36. 42.0 Gemini 3.6 Flash
  37. 41.0 Claude Opus 4.7
  38. 41.0 Grok 4.7
  39. 40.0 Gemini 3 Flash
  40. 40.0 Doubao Seed Evolving
  41. 37.0 o3
  42. 37.0 Kimi K3
  43. 37.0 GLM-5.3
  44. 37.0 Qwen3.8 Max (0902)
  45. 37.0 Muse Spark 1.3
  46. 36.0 GPT-5
  47. 36.0 GPT-6 Luna
  48. 35.0 Grok 4.5
  49. 34.0 Muse Spark 1.1
  50. 33.0 Grok 4.6
  51. 33.0 Mistral Large 4 Preview
  52. 32.0 DeepSeek V4 Flash 0731
  53. 29.0 Grok 4.20
  54. 29.0 Mistral Large 4
  55. 27.0 Qwen3.7-Max
  56. 27.0 Muse Spark 1.2
  57. 26.0 Doubao Seed 2.1 Turbo
  58. 24.0 GPT-5 mini
  59. 23.0 Claude Sonnet 4.6
  60. 23.0 GPT-5.4 Mini
  61. 22.0 o4-mini
  62. 22.0 GPT-5 nano
  63. 22.0 GLM-5.2
  64. 21.0 GLM-4.7
  65. 21.0 Claude Opus 4.6
  66. 21.0 Doubao Seed 2.1 Pro
  67. 21.0 Inkling
  68. 21.0 Gemini 3.5 Flash-Lite
  69. 21.0 GLM-5.3-Flash
  70. 20.0 Doubao-Seed-2.0 Pro
  71. 20.0 GLM-5.1
  72. 20.0 Kimi K2.6
  73. 20.0 Doubao Seed 2.0 Lite
  74. 19.0 DeepSeek-V4-Pro
  75. 18.0 Grok 4.3 Beta
  76. 17.0 MiniMax-M3
  77. 16.0 Qwen 3.6 Plus
  78. 15.0 o3-mini
  79. 14.0 Claude Opus 4.5
  80. 14.0 Qwen 3.5 Flash (hosted 35B-A3B)
  81. 14.0 Kimi K2.7 Code
  82. 12.0 Qwen 3.6 Flash
  83. 11.0 o1
  84. 11.0 GPT-4.1
  85. 10.0 Claude Sonnet 4.5
  86. 9.00 GPT-4o
  87. 9.00 Qwen 3.5 Plus (hosted 397B-A17B)
  88. 8.00 Claude Haiku 4.5
  89. 5.00 Doubao-Seed-1.6
  90. 4.00 Claude 3 Opus
  91. 3.00 GPT-4o mini
  92. 1.00 GPT-4.1 mini
  93. 0.0 GPT-4.1 nano
  1. 1 Claude Fable 5.2 max 79%
  2. 2 Gemini 4 Carbon high 78%
  3. 3 GPT-6 Astra pro, max 76%
  4. 4 GPT-5.6 Sol pro, max 68%
  5. 5 GPT-6.1 Sol max 68%
  6. 6 Claude Opus 5.5 max 67%
  7. 7 Gemini 3.7 Flash high 66%
  8. 8 GPT-5.5 xhigh 63%
  9. 9 GPT-5.5 Pro xhigh 62%
  10. 10 GPT-5.2 Pro xhigh 60%
  11. 11 GPT-5.4 Pro xhigh 59%
  12. 12 Claude Sonnet 5.5 max 59%
  13. 13 Gemini 4 Argon high 59%
  14. 14 Gemini 3.8 Flash high 58%
  15. 15 GPT-6 Sol max 56%
  16. 16 Claude Haiku 5.5 max 55%
  17. 17 Claude Opus 4.8 max 52%
  18. 18 Gemini 3.1 Pro high 51%
  19. 19 GPT-5 Pro high 48%
  20. 20 GPT-5.2 xhigh 48%
  21. 21 GPT-5.4 xhigh 48%
  22. 22 Claude Fable 5.1 max 48%
  23. 23 DeepSeek V4.1 Flash max 48%
  24. 24 Claude Fable 5 max 47%
  25. 25 GPT-5.6 Terra max 47%
  26. 26 Qwen 3.8 Max xhigh 46%
  27. 27 DeepSeek V4 Pro 0813 max 46%
  28. 28 DeepSeek V4.1 Flash 1017 max 46%
  29. 29 GPT-5.4 Nano xhigh 45%
  30. 30 Gemini 3.5 Flash high 45%
  31. 31 Claude Sonnet 5 max 45%
  32. 32 Kimi K3.1 Pro Together, max 45%
  33. 33 GPT-5.6 Luna max 43%
  34. 34 Claude Opus 5 max 43%
  35. 35 GPT-5.1 high 42%
  36. 36 Gemini 3.6 Flash low 42%
  37. 37 Claude Opus 4.7 max 41%
  38. 38 Grok 4.7 xhigh 41%
  39. 39 Gemini 3 Flash high 40%
  40. 40 Doubao Seed Evolving Nov 2026, high 40%
  41. 41 o3 medium 37%
  42. 42 Kimi K3 high 37%
  43. 43 GLM-5.3 max 37%
  44. 44 Qwen3.8 Max (0902) xhigh 37%
  45. 45 Muse Spark 1.3 max 37%
  46. 46 GPT-5 high 36%
  47. 47 GPT-6 Luna max 36%
  48. 48 Grok 4.5 high 35%
  49. 49 Muse Spark 1.1 xhigh 34%
  50. 50 Grok 4.6 xhigh 33%
  51. 51 Mistral Large 4 Preview high 33%
  52. 52 DeepSeek V4 Flash 0731 max 32%
  53. 53 Grok 4.20 29%
  54. 54 Mistral Large 4 high 29%
  55. 55 Qwen3.7-Max 27%
  56. 56 Muse Spark 1.2 xhigh 27%
  57. 57 Doubao Seed 2.1 Turbo high 26%
  58. 58 GPT-5 mini high 24%
  59. 59 Claude Sonnet 4.6 max 23%
  60. 60 GPT-5.4 Mini high 23%
  61. 61 o4-mini medium 22%
  62. 62 GPT-5 nano high 22%
  63. 63 GLM-5.2 max 22%
  64. 64 GLM-4.7 21%
  65. 65 Claude Opus 4.6 max 21%
  66. 66 Doubao Seed 2.1 Pro high 21%
  67. 67 Inkling xhigh 21%
  68. 68 Gemini 3.5 Flash-Lite low 21%
  69. 69 GLM-5.3-Flash max 21%
  70. 70 Doubao-Seed-2.0 Pro high 20%
  71. 71 GLM-5.1 20%
  72. 72 Kimi K2.6 20%
  73. 73 Doubao Seed 2.0 Lite Apr 2026, high 20%
  74. 74 DeepSeek-V4-Pro max 19%
  75. 75 Grok 4.3 Beta high 18%
  76. 76 MiniMax-M3 17%
  77. 77 Qwen 3.6 Plus 2026-04-02 16%
  78. 78 o3-mini high 15%
  79. 79 Claude Opus 4.5 59k thinking 14%
  80. 80 Qwen 3.5 Flash (hosted 35B-A3B) 14%
  81. 81 Kimi K2.7 Code 14%
  82. 82 Qwen 3.6 Flash 2026-04-16 12%
  83. 83 o1 low 11%
  84. 84 GPT-4.1 11%
  85. 85 Claude Sonnet 4.5 59k thinking 10%
  86. 86 GPT-4o Aug 2024 9.00%
  87. 87 Qwen 3.5 Plus (hosted 397B-A17B) 9.00%
  88. 88 Claude Haiku 4.5 59k thinking 8.00%
  89. 89 Doubao-Seed-1.6 thinking 5.00%
  90. 90 Claude 3 Opus 4.00%
  91. 91 GPT-4o mini 3.00%
  92. 92 GPT-4.1 mini 1.00%
  93. 93 GPT-4.1 nano 0%

Source: Epoch AI runs (CC BY 4.0)

#ModelScoreSettingDateSource
1 Claude Fable 5.2 79% max Nov 2026 Epoch AI run
2 Gemini 4 Carbon 78% high Oct 2026 Epoch AI run
3 GPT-6 Astra 76% pro, max Sep 2026 Epoch AI run
4 GPT-5.6 Sol 68% pro, max Jul 2026 Epoch AI run
5 GPT-6.1 Sol 68% max Sep 2026 Epoch AI run
6 Claude Opus 5.5 67% max Sep 2026 Epoch AI run
7 Gemini 3.7 Flash 66% high Aug 2026 Epoch AI run
8 GPT-5.5 63% xhigh Apr 2026 Epoch AI run
9 GPT-5.5 Pro 62% xhigh Apr 2026 Epoch AI run
10 GPT-5.2 Pro 60% xhigh Dec 2025 Epoch AI run
11 GPT-5.4 Pro 59% xhigh Mar 2026 Epoch AI run
12 Claude Sonnet 5.5 59% max Sep 2026 Epoch AI run
13 Gemini 4 Argon 59% high Sep 2026 Epoch AI run
14 Gemini 3.8 Flash 58% high Sep 2026 Epoch AI run
15 GPT-6 Sol 56% max Sep 2026 Epoch AI run
16 Claude Haiku 5.5 55% max Oct 2026 Epoch AI run
17 Claude Opus 4.8 52% max May 2026 Epoch AI run
18 Gemini 3.1 Pro 51% high Feb 2026 Epoch AI run
19 GPT-5 Pro 48% high Oct 2025 Epoch AI run
20 GPT-5.2 48% xhigh Dec 2025 Epoch AI run
21 GPT-5.4 48% xhigh Mar 2026 Epoch AI run
22 Claude Fable 5.1 48% max Sep 2026 Epoch AI run
23 DeepSeek V4.1 Flash 48% max Sep 2026 Epoch AI run
24 Claude Fable 5 47% max Jun 2026 Epoch AI run
25 GPT-5.6 Terra 47% max Jul 2026 Epoch AI run
26 Qwen 3.8 Max 46% xhigh Aug 2026 Epoch AI run
27 DeepSeek V4 Pro 0813 46% max Aug 2026 Epoch AI run
28 DeepSeek V4.1 Flash 1017 46% max Oct 2026 Epoch AI run
29 GPT-5.4 Nano 45% xhigh Mar 2026 Epoch AI run
30 Gemini 3.5 Flash 45% high May 2026 Epoch AI run
31 Claude Sonnet 5 45% max Jun 2026 Epoch AI run
32 Kimi K3.1 Pro 45% Together, max Oct 2026 Epoch AI run
33 GPT-5.6 Luna 43% max Jul 2026 Epoch AI run
34 Claude Opus 5 43% max Jul 2026 Epoch AI run
35 GPT-5.1 42% high Nov 2025 Epoch AI run
36 Gemini 3.6 Flash 42% low Jul 2026 Epoch AI run
37 Claude Opus 4.7 41% max Apr 2026 Epoch AI run
38 Grok 4.7 41% xhigh Sep 2026 Epoch AI run
39 Gemini 3 Flash 40% high Dec 2025 Epoch AI run
40 Doubao Seed Evolving Nov 2026 40% high Nov 2026 Epoch AI run
41 o3 37% medium Apr 2025 Epoch AI run
42 Kimi K3 37% high Jul 2026 Epoch AI run
43 GLM-5.3 37% max Aug 2026 Epoch AI run
44 Qwen3.8 Max (0902) 37% xhigh Sep 2026 Epoch AI run
45 Muse Spark 1.3 37% max Sep 2026 Epoch AI run
46 GPT-5 36% high Aug 2025 Epoch AI run
47 GPT-6 Luna 36% max Sep 2026 Epoch AI run
48 Grok 4.5 35% high Jul 2026 Epoch AI run
49 Muse Spark 1.1 34% xhigh Jul 2026 Epoch AI run
50 Grok 4.6 33% xhigh Aug 2026 Epoch AI run
51 Mistral Large 4 Preview 33% high Oct 2026 Epoch AI run
52 DeepSeek V4 Flash 0731 32% max Jul 2026 Epoch AI run
53 Grok 4.20 29% Feb 2026 Epoch AI run
54 Mistral Large 4 29% high Oct 2026 Epoch AI run
55 Qwen3.7-Max 27% May 2026 Epoch AI run
56 Muse Spark 1.2 27% xhigh Aug 2026 Epoch AI run
57 Doubao Seed 2.1 Turbo 26% high Jun 2026 Epoch AI run
58 GPT-5 mini 24% high Aug 2025 Epoch AI run
59 Claude Sonnet 4.6 23% max Feb 2026 Epoch AI run
60 GPT-5.4 Mini 23% high Mar 2026 Epoch AI run
61 o4-mini 22% medium Apr 2025 Epoch AI run
62 GPT-5 nano 22% high Aug 2025 Epoch AI run
63 GLM-5.2 22% max Jun 2026 Epoch AI run
64 GLM-4.7 21% Dec 2025 Epoch AI run
65 Claude Opus 4.6 21% max Feb 2026 Epoch AI run
66 Doubao Seed 2.1 Pro 21% high Jun 2026 Epoch AI run
67 Inkling 21% xhigh Jul 2026 Epoch AI run
68 Gemini 3.5 Flash-Lite 21% low Jul 2026 Epoch AI run
69 GLM-5.3-Flash 21% max Aug 2026 Epoch AI run
70 Doubao-Seed-2.0 Pro 20% high Feb 2026 Epoch AI run
71 GLM-5.1 20% Apr 2026 Epoch AI run
72 Kimi K2.6 20% Apr 2026 Epoch AI run
73 Doubao Seed 2.0 Lite Apr 2026 20% high Apr 2026 Epoch AI run
74 DeepSeek-V4-Pro 19% max Apr 2026 Epoch AI run
75 Grok 4.3 Beta 18% high Apr 2026 Epoch AI run
76 MiniMax-M3 17% Jun 2026 Epoch AI run
77 Qwen 3.6 Plus 16% 2026-04-02 Mar 2026 Epoch AI run
78 o3-mini 15% high Jan 2025 Epoch AI run
79 Claude Opus 4.5 14% 59k thinking Nov 2025 Epoch AI run
80 Qwen 3.5 Flash (hosted 35B-A3B) 14% Feb 2026 Epoch AI run
81 Kimi K2.7 Code 14% Jun 2026 Epoch AI run
82 Qwen 3.6 Flash 12% 2026-04-16 Apr 2026 Epoch AI run
83 o1 11% low Dec 2024 Epoch AI run
84 GPT-4.1 11% Apr 2025 Epoch AI run
85 Claude Sonnet 4.5 10% 59k thinking Sep 2025 Epoch AI run
86 GPT-4o Aug 2024 9.00% Aug 2024 Epoch AI run
87 Qwen 3.5 Plus (hosted 397B-A17B) 9.00% Feb 2026 Epoch AI run
88 Claude Haiku 4.5 8.00% 59k thinking Oct 2025 Epoch AI run
89 Doubao-Seed-1.6 5.00% thinking Jun 2025 Epoch AI run
90 Claude 3 Opus 4.00% Feb 2024 Epoch AI run
91 GPT-4o mini 3.00% Jul 2024 Epoch AI run
92 GPT-4.1 mini 1.00% Apr 2025 Epoch AI run
93 GPT-4.1 nano 0% Apr 2025 Epoch AI run

7 models have held number one in Epoch AI’s runs since its release in December 2025

Lab claims count from the day the card came out, independent runs from the model’s release.

OpenAI Google DeepMind Anthropic
Epoch AI runs 7 leaders
Claude Fable 5.2 79%

Source: Epoch AI runs (CC BY 4.0)

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Run by Epoch AI Best so far
0% 20% 40% 60% 80% 100% 2024 2025 2026 2027 Near the ceiling (85%) Random guessing Claude Fable 5.2 GPT-5.6 Sol GPT-5.5 GPT-5 Pro o3 o3-mini GPT-4o 0% 20% 40% 60% 80% 100% 2024 2025 2026 2027 Near the ceiling (85%) Random guessing Claude Fable 5.2 GPT-5.6 Sol

Source: Epoch AI runs (CC BY 4.0)

No lab model card reports it yet

We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark.

More in Games and long tasks

All 82 in Games and long tasks
  • ARC-AGI-3 In use Interactive game environments where agents must explore, infer goals and rules, and act efficiently without instructions. GPT-6 Astra 99.95%
  • Orak Fading LLM agents play 12 popular video games across most genres through an MCP interface. Each game scores in-game progress or outcomes, and the board ranks models by their average rank across the games. Gemini 2.5 Pro 3.90
  • MineDojo Fading Agents complete open-ended tasks in Minecraft, such as mining blocks and fighting mobs, from a large simulation suite. Programmatic checks score many tasks. Seed2.0 Pro 49%
  • TextQuests Only in system cards Agents play 25 classic interactive fiction games. The score is harmful actions per 100 steps, where lower is better. The board also reports game progress. Llama 4 Scout 7.00
  • GameWorld Only in system cards Multimodal game agents play 34 browser games across 170 tasks. Outcomes are checked from the game state. GPT-6.1 Terra 76.8%
  • Swarm Bench Only in system cards Models orchestrate agent swarms on complex tasks that benefit from coordinated decomposition and parallel execution. Kimi K3 76.3%