Benchmark Almanac

ACCoRD

Claude Fable 5.2 89.4 leads since November 2026
Not in model cards

Models state probabilities and preferences on conceptual questions, and the benchmark checks 567 expert-approved logical consistency constraints such as conjunction rules and order invariance.

A component of the Conceptual Reasoning Index. The full dataset has 13,853 model-generated constraints in 18 types. No paper yet.

Released
August 2026
Kind
Other
Level
Expert
Scored
Checked automatically
Answers
Prediction
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
154
Entries on its own leaderboard, 152 models we track

Claude Fable 5.2 leads with 89.4

Best score per entry from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders.

154 entries. Scroll sideways for the rest.

  1. 89.4 Claude Fable 5.2
  2. 88.4 Claude Opus 5.5
  3. 85.7 Claude Sonnet 5.5
  4. 85.0 Claude Fable 5
  5. 83.8 Claude Opus 5
  6. 83.7 Claude Fable 5.1
  7. 82.4 GPT-6 Sol
  8. 81.7 GPT-6 Astra
  9. 81.5 Claude Haiku 5.5
  10. 80.9 GPT-6.1 Sol
  11. 80.5 GPT-5.5
  12. 80.4 Claude Opus 4.6
  13. 80.0 GPT-5.4
  14. 79.8 Claude Opus 4.7
  15. 79.3 GPT-5.2
  16. 79.1 Claude Opus 4.5
  17. 78.9 GPT-5.6 Sol Pro
  18. 78.4 Muse Spark 1.1
  19. 78.2 Gemini 3.8 Flash
  20. 77.5 Kimi K3.1 Pro
  21. 77.4 Claude Opus 4.8
  22. 77.4 Muse Spark 1.2
  23. 77.3 Gemini 4 Argon
  24. 77.0 Gemini 3.7 Flash
  25. 76.9 Claude Sonnet 5
  26. 76.2 Muse Wonder
  27. 75.9 GPT-5.6 Sol
  28. 75.8 GPT-5.5 Pro
  29. 75.1 Muse Spark 1.3
  30. 74.7 GPT-5.6 Terra
  31. 74.0 Grok 4.7
  32. 73.8 o3-pro
  33. 73.5 Gemini 3.1 Pro
  34. 73.4 Claude Sonnet 4.6
  35. 73.3 Qwen 3.8 27B
  36. 73.1 GPT-5
  37. 72.1 GPT-6 Luna
  38. 72.0 Muse Glimmer
  39. 71.6 Grok 4.5
  40. 71.6 Qwen3.8-2.4T-A95B
  41. 71.6 Grok 4.6
  42. 71.5 GPT-5 mini
  43. 71.4 Qwen 3.8 Max
  44. 71.4 o3
  45. 71.4 Kimi K2.6
  46. 71.2 Inkling
  47. 71.2 GPT-5.1
  48. 70.9 DeepSeek-V4-Pro
  49. 70.9 GLM-5.3
  50. 70.8 Qwen 3.6 Max (Preview)
  51. 70.7 Claude Sonnet 4.5
  52. 70.5 GPT-5.4 Mini
  53. 70.5 Kimi K3
  54. 70.2 DeepSeek V4 Pro 0813
  55. 70.0 Gemini 3.6 Flash
  56. 69.6 Grok 4.20 Multi-agent
  57. 68.8 Grok 4.20
  58. 68.6 Gemini 3 Flash
  59. 68.1 DeepSeek V4.1 Flash
  60. 68.0 Gemini 3.5 Flash-Lite
  61. 67.5 gpt-oss-120b
  62. 67.5 GPT-5.6 Luna
  63. 67.3 GPT-5.4 Nano
  64. 66.7 Claude Haiku 4.5
  65. 66.6 Gemini 3.5 Flash
  66. 66.4 Tencent Hy3
  67. 66.2 Qwen Plus
  68. 66.0 GLM-5.2
  69. 66.0 GPT-5.2 Instant
  70. 65.9 MiniMax-M3
  71. 65.8 Claude Opus 4
  72. 65.7 Mistral Large 4
  73. 65.3 Qwen3.7-Max
  74. 65.3 Grok 4.3
  75. 64.6 DeepSeek-V3.2
  76. 64.6 Nemotron 3 Ultra
  77. 64.2 o1
  78. 64.1 Gemini 3.1 Flash-Lite
  79. 63.8 o4-mini
  80. 63.5 Claude Opus 4.1
  81. 62.8 Qwen3-Max
  82. 62.7 Gemini 2.5 Pro
  83. 62.0 Gemma 4 31B IT
  84. 61.9 gpt-oss-20b
  85. 61.8 DeepSeek-V3.2-Exp
  86. 61.7 DeepSeek V4 Flash 0731
  87. 61.5 Qwen 3.6 Flash
  88. 61.1 GPT-5.3 Instant
  89. 60.7 Qwen3.7-Plus
  90. 60.5 Qwen3.5-27B
  91. 59.6 Qwen3-Next-80B-A3B
  92. 59.3 Qwen 3.6 Plus
  93. 58.9 Qwen3.6 27B
  94. 58.8 GPT-5 nano
  95. 58.7 Qwen3 235B-A22B (2507)
  96. 58.2 MiMo-V2.5-Pro
  97. 58.0 Qwen3.5-122B-A10B
  98. 57.3 DeepSeek-V4-Flash
  99. 57.1 Qwen3-235B-A22B-Thinking
  100. 57.0 Qwen3-30B-A3B-Thinking
  101. 57.0 Claude Sonnet 4
  102. 56.5 o3-mini
  103. 56.1 Qwen 3.5 Plus (hosted 397B-A17B)
  104. 56.0 GPT-4.1
  105. 55.6 Qwen3.5 397B-A17B
  106. 55.5 DeepSeek-V3.1
  107. 55.3 Qwen3.5-35B-A3B
  108. 55.2 DeepSeek-V3.1-Terminus
  109. 54.7 Nemotron 3.5 Lightning 30B-A3B
  110. 54.3 Gemma 4 26B A4B
  111. 53.0 Qwen 3.6 35B-A3B
  112. 52.5 Qwen 3.5 Flash (hosted 35B-A3B)
  113. 50.2 Qwen3.5-9B
  114. 50.2 Llama 4 Maverick
  115. 49.7 Llama 4 Scout
  116. 48.7 Mistral Large 3
  117. 48.4 Qwen3-14B
  118. 48.2 Qwen3-32B
  119. 46.8 Qwen3-235B-A22B
  120. 45.6 Mistral Medium 3.5
  121. 45.5 Llama 3.3 70B
  122. 45.4 Qwen3-30B-A3B-Instruct
  123. 45.4 Gemini 2.5 Flash
  124. 45.0 Qwen3-30B-A3B
  125. 44.8 GPT-4o
  126. 43.6 GPT-4.1 mini
  127. 43.4 DeepSeek-V3
  128. 43.2 Llama 3.1-70B
  129. 42.5 Qwen2.5-72B
  130. 42.2 GPT-4 Turbo
  131. 41.8 Qwen3-8B
  132. 41.6 DeepSeek Chat
  133. 41.2 Gemma 3n E4B
  134. 40.5 Gemma 3 27B
  135. 39.2 Mistral Medium 3.1
  136. 39.0 Mistral Small 4
  137. 38.3 Gemma 3 12B
  138. 37.3 GPT-4
  139. 37.3 Cohere Command A
  140. 33.7 Llama 3.1-8B
  141. 33.5 Ministral 3 8B Instruct
  142. 33.4 Gemini 2.5 Flash-Lite
  143. 33.1 Claude 3 Opus
  144. 32.8 Ministral 3 14B
  145. 32.6 GPT-4o mini
  146. 27.3 Qwen2.5-7B
  147. 24.3 GPT-4.1 nano
  148. 23.8 Gemma 3 4B
  149. 20.2 GPT-3.5 Turbo
  150. 18.0 Command R+
  151. 14.7 Claude 3 Haiku
  152. 14.6 Ministral 3B
  153. 13.0 c4ai-command-r-08-2024
  154. 9.94 Command R7B
  1. 1 Claude Fable 5.2 xhigh 89.4
  2. 2 Claude Opus 5.5 high 88.4
  3. 3 Claude Sonnet 5.5 xhigh 85.7
  4. 4 Claude Fable 5 max 85.0
  5. 5 Claude Opus 5 low 83.8
  6. 6 Claude Fable 5.1 medium 83.7
  7. 7 GPT-6 Sol 82.4
  8. 8 GPT-6 Astra max 81.7
  9. 9 Claude Haiku 5.5 max 81.5
  10. 10 GPT-6.1 Sol 80.9
  11. 11 GPT-5.5 80.5
  12. 12 Claude Opus 4.6 80.4
  13. 13 GPT-5.4 80.0
  14. 14 Claude Opus 4.7 79.8
  15. 15 GPT-5.2 79.3
  16. 16 Claude Opus 4.5 79.1
  17. 17 GPT-5.6 Sol Pro 78.9
  18. 18 Muse Spark 1.1 78.4
  19. 19 Gemini 3.8 Flash medium 78.2
  20. 20 Kimi K3.1 Pro 77.5
  21. 21 Claude Opus 4.8 77.4
  22. 22 Muse Spark 1.2 77.4
  23. 23 Gemini 4 Argon low 77.3
  24. 24 Gemini 3.7 Flash high 77.0
  25. 25 Claude Sonnet 5 xhigh 76.9
  26. 26 Muse Wonder 76.2
  27. 27 GPT-5.6 Sol max 75.9
  28. 28 GPT-5.5 Pro 75.8
  29. 29 Muse Spark 1.3 75.1
  30. 30 GPT-5.6 Terra max 74.7
  31. 31 Grok 4.7 74.0
  32. 32 o3-pro 73.8
  33. 33 Gemini 3.1 Pro medium 73.5
  34. 34 Claude Sonnet 4.6 73.4
  35. 35 Qwen 3.8 27B 73.3
  36. 36 GPT-5 73.1
  37. 37 GPT-6 Luna 72.1
  38. 38 Muse Glimmer 72.0
  39. 39 Grok 4.5 71.6
  40. 40 Qwen3.8-2.4T-A95B 71.6
  41. 41 Grok 4.6 71.6
  42. 42 GPT-5 mini 71.5
  43. 43 Qwen 3.8 Max 71.4
  44. 44 o3 71.4
  45. 45 Kimi K2.6 71.4
  46. 46 Inkling 71.2
  47. 47 GPT-5.1 71.2
  48. 48 DeepSeek-V4-Pro 70.9
  49. 49 GLM-5.3 70.9
  50. 50 Qwen 3.6 Max (Preview) 70.8
  51. 51 Claude Sonnet 4.5 70.7
  52. 52 GPT-5.4 Mini 70.5
  53. 53 Kimi K3 70.5
  54. 54 DeepSeek V4 Pro 0813 70.2
  55. 55 Gemini 3.6 Flash 70.0
  56. 56 Grok 4.20 Multi-agent 69.6
  57. 57 Grok 4.20 68.8
  58. 58 Gemini 3 Flash 68.6
  59. 59 DeepSeek V4.1 Flash 68.1
  60. 60 Gemini 3.5 Flash-Lite 68.0
  61. 61 gpt-oss-120b 67.5
  62. 62 GPT-5.6 Luna high 67.5
  63. 63 GPT-5.4 Nano 67.3
  64. 64 Claude Haiku 4.5 66.7
  65. 65 Gemini 3.5 Flash 66.6
  66. 66 Tencent Hy3 66.4
  67. 67 Qwen Plus Apr 2025 66.2
  68. 68 GLM-5.2 66.0
  69. 69 GPT-5.2 Instant 66.0
  70. 70 MiniMax-M3 65.9
  71. 71 Claude Opus 4 65.8
  72. 72 Mistral Large 4 65.7
  73. 73 Qwen3.7-Max 65.3
  74. 74 Grok 4.3 65.3
  75. 75 DeepSeek-V3.2 64.6
  76. 76 Nemotron 3 Ultra 64.6
  77. 77 o1 64.2
  78. 78 Gemini 3.1 Flash-Lite 64.1
  79. 79 o4-mini 63.8
  80. 80 Claude Opus 4.1 63.5
  81. 81 Qwen3-Max thinking 62.8
  82. 82 Gemini 2.5 Pro Jun 2025 62.7
  83. 83 Gemma 4 31B IT 62.0
  84. 84 gpt-oss-20b 61.9
  85. 85 DeepSeek-V3.2-Exp 61.8
  86. 86 DeepSeek V4 Flash 0731 61.7
  87. 87 Qwen 3.6 Flash 61.5
  88. 88 GPT-5.3 Instant 61.1
  89. 89 Qwen3.7-Plus 60.7
  90. 90 Qwen3.5-27B 60.5
  91. 91 Qwen3-Next-80B-A3B thinking 59.6
  92. 92 Qwen 3.6 Plus 59.3
  93. 93 Qwen3.6 27B 58.9
  94. 94 GPT-5 nano 58.8
  95. 95 Qwen3 235B-A22B (2507) 58.7
  96. 96 MiMo-V2.5-Pro 58.2
  97. 97 Qwen3.5-122B-A10B 58.0
  98. 98 DeepSeek-V4-Flash 57.3
  99. 99 Qwen3-235B-A22B-Thinking Jul 2025 57.1
  100. 100 Qwen3-30B-A3B-Thinking Jul 2025 57.0
  101. 101 Claude Sonnet 4 57.0
  102. 102 o3-mini 56.5
  103. 103 Qwen 3.5 Plus (hosted 397B-A17B) 56.1
  104. 104 GPT-4.1 56.0
  105. 105 Qwen3.5 397B-A17B 55.6
  106. 106 DeepSeek-V3.1 non-thinking 55.5
  107. 107 Qwen3.5-35B-A3B 55.3
  108. 108 DeepSeek-V3.1-Terminus 55.2
  109. 109 Nemotron 3.5 Lightning 30B-A3B 54.7
  110. 110 Gemma 4 26B A4B 54.3
  111. 111 Qwen 3.6 35B-A3B 53.0
  112. 112 Qwen 3.5 Flash (hosted 35B-A3B) 52.5
  113. 113 Qwen3.5-9B 50.2
  114. 114 Llama 4 Maverick 50.2
  115. 115 Llama 4 Scout 49.7
  116. 116 Mistral Large 3 48.7
  117. 117 Qwen3-14B 48.4
  118. 118 Qwen3-32B 48.2
  119. 119 Qwen3-235B-A22B 46.8
  120. 120 Mistral Medium 3.5 45.6
  121. 121 Llama 3.3 70B 45.5
  122. 122 Qwen3-30B-A3B-Instruct Jul 2025 45.4
  123. 123 Gemini 2.5 Flash Jun 2025 45.4
  124. 124 Qwen3-30B-A3B 45.0
  125. 125 GPT-4o May 2024 44.8
  126. 126 GPT-4.1 mini 43.6
  127. 127 DeepSeek-V3 Mar 2025 43.4
  128. 128 Llama 3.1-70B 43.2
  129. 129 Qwen2.5-72B 42.5
  130. 130 GPT-4 Turbo Apr 2024 42.2
  131. 131 Qwen3-8B 41.8
  132. 132 DeepSeek Chat 41.6
  133. 133 Gemma 3n E4B 41.2
  134. 134 Gemma 3 27B 40.5
  135. 135 Mistral Medium 3.1 39.2
  136. 136 Mistral Small 4 39.0
  137. 137 Gemma 3 12B 38.3
  138. 138 GPT-4 Jun 2023 37.3
  139. 139 Cohere Command A 37.3
  140. 140 Llama 3.1-8B 33.7
  141. 141 Ministral 3 8B Instruct 33.5
  142. 142 Gemini 2.5 Flash-Lite Jun 2025 33.4
  143. 143 Claude 3 Opus 33.1
  144. 144 Ministral 3 14B 32.8
  145. 145 GPT-4o mini 32.6
  146. 146 Qwen2.5-7B 27.3
  147. 147 GPT-4.1 nano 24.3
  148. 148 Gemma 3 4B 23.8
  149. 149 GPT-3.5 Turbo Nov 2023 20.2
  150. 150 Command R+ 18.0
  151. 151 Claude 3 Haiku 14.7
  152. 152 Ministral 3B 14.6
  153. 153 c4ai-command-r-08-2024 13.0
  154. 154 Command R7B 9.94

Source: Official ACCoRD leaderboard

#EntryScoreSettingDateSource
1 Claude Fable 5.2 89.4 xhigh Nov 2026 Official leaderboard
2 Claude Opus 5.5 88.4 high Sep 2026 Official leaderboard
3 Claude Sonnet 5.5 85.7 xhigh Sep 2026 Official leaderboard
4 Claude Fable 5 85.0 max Jun 2026 Official leaderboard
5 Claude Opus 5 83.8 low Jul 2026 Official leaderboard
6 Claude Fable 5.1 83.7 medium Sep 2026 Official leaderboard
7 GPT-6 Sol 82.4 Sep 2026 Official leaderboard
8 GPT-6 Astra 81.7 max Sep 2026 Official leaderboard
9 Claude Haiku 5.5 81.5 max Oct 2026 Official leaderboard
10 GPT-6.1 Sol 80.9 Sep 2026 Official leaderboard
11 GPT-5.5 80.5 Apr 2026 Official leaderboard
12 Claude Opus 4.6 80.4 Feb 2026 Official leaderboard
13 GPT-5.4 80.0 Mar 2026 Official leaderboard
14 Claude Opus 4.7 79.8 Apr 2026 Official leaderboard
15 GPT-5.2 79.3 Dec 2025 Official leaderboard
16 Claude Opus 4.5 79.1 Nov 2025 Official leaderboard
17 GPT-5.6 Sol Pro 78.9 Jul 2026 Official leaderboard
18 Muse Spark 1.1 78.4 Jul 2026 Official leaderboard
19 Gemini 3.8 Flash 78.2 medium Sep 2026 Official leaderboard
20 Kimi K3.1 Pro 77.5 Oct 2026 Official leaderboard
21 Claude Opus 4.8 77.4 May 2026 Official leaderboard
22 Muse Spark 1.2 77.4 Aug 2026 Official leaderboard
23 Gemini 4 Argon 77.3 low Sep 2026 Official leaderboard
24 Gemini 3.7 Flash 77.0 high Aug 2026 Official leaderboard
25 Claude Sonnet 5 76.9 xhigh Jun 2026 Official leaderboard
26 Muse Wonder 76.2 Oct 2026 Official leaderboard
27 GPT-5.6 Sol 75.9 max Jul 2026 Official leaderboard
28 GPT-5.5 Pro 75.8 Apr 2026 Official leaderboard
29 Muse Spark 1.3 75.1 Sep 2026 Official leaderboard
30 GPT-5.6 Terra 74.7 max Jul 2026 Official leaderboard
31 Grok 4.7 74.0 Sep 2026 Official leaderboard
32 o3-pro 73.8 Jun 2025 Official leaderboard
33 Gemini 3.1 Pro 73.5 medium Feb 2026 Official leaderboard
34 Claude Sonnet 4.6 73.4 Feb 2026 Official leaderboard
35 Qwen 3.8 27B 73.3 Aug 2026 Official leaderboard
36 GPT-5 73.1 Aug 2025 Official leaderboard
37 GPT-6 Luna 72.1 Sep 2026 Official leaderboard
38 Muse Glimmer 72.0 Aug 2026 Official leaderboard
39 Grok 4.5 71.6 Jul 2026 Official leaderboard
40 Qwen3.8-2.4T-A95B 71.6 Aug 2026 Official leaderboard
41 Grok 4.6 71.6 Aug 2026 Official leaderboard
42 GPT-5 mini 71.5 Aug 2025 Official leaderboard
43 Qwen 3.8 Max 71.4 Aug 2026 Official leaderboard
44 o3 71.4 Apr 2025 Official leaderboard
45 Kimi K2.6 71.4 Apr 2026 Official leaderboard
46 Inkling 71.2 Jul 2026 Official leaderboard
47 GPT-5.1 71.2 Nov 2025 Official leaderboard
48 DeepSeek-V4-Pro 70.9 Apr 2026 Official leaderboard
49 GLM-5.3 70.9 Aug 2026 Official leaderboard
50 Qwen 3.6 Max (Preview) 70.8 Apr 2026 Official leaderboard
51 Claude Sonnet 4.5 70.7 Sep 2025 Official leaderboard
52 GPT-5.4 Mini 70.5 Mar 2026 Official leaderboard
53 Kimi K3 70.5 Jul 2026 Official leaderboard
54 DeepSeek V4 Pro 0813 70.2 Aug 2026 Official leaderboard
55 Gemini 3.6 Flash 70.0 Jul 2026 Official leaderboard
56 Grok 4.20 Multi-agent 69.6 Mar 2026 Official leaderboard
57 Grok 4.20 68.8 Feb 2026 Official leaderboard
58 Gemini 3 Flash 68.6 Dec 2025 Official leaderboard
59 DeepSeek V4.1 Flash 68.1 Sep 2026 Official leaderboard
60 Gemini 3.5 Flash-Lite 68.0 Jul 2026 Official leaderboard
61 gpt-oss-120b 67.5 Aug 2025 Official leaderboard
62 GPT-5.6 Luna 67.5 high Jul 2026 Official leaderboard
63 GPT-5.4 Nano 67.3 Mar 2026 Official leaderboard
64 Claude Haiku 4.5 66.7 Oct 2025 Official leaderboard
65 Gemini 3.5 Flash 66.6 May 2026 Official leaderboard
66 Tencent Hy3 66.4 Jul 2026 Official leaderboard
67 Qwen Plus Apr 2025 66.2 Apr 2025 Official leaderboard
68 GLM-5.2 66.0 Jun 2026 Official leaderboard
69 GPT-5.2 Instant 66.0 Dec 2025 Official leaderboard
70 MiniMax-M3 65.9 Jun 2026 Official leaderboard
71 Claude Opus 4 65.8 May 2025 Official leaderboard
72 Mistral Large 4 65.7 Oct 2026 Official leaderboard
73 Qwen3.7-Max 65.3 May 2026 Official leaderboard
74 Grok 4.3 65.3 Apr 2026 Official leaderboard
75 DeepSeek-V3.2 64.6 Dec 2025 Official leaderboard
76 Nemotron 3 Ultra 64.6 Jun 2026 Official leaderboard
77 o1 64.2 Dec 2024 Official leaderboard
78 Gemini 3.1 Flash-Lite 64.1 Mar 2026 Official leaderboard
79 o4-mini 63.8 Apr 2025 Official leaderboard
80 Claude Opus 4.1 63.5 Aug 2025 Official leaderboard
81 Qwen3-Max 62.8 thinking Sep 2025 Official leaderboard
82 Gemini 2.5 Pro Jun 2025 62.7 Jun 2025 Official leaderboard
83 Gemma 4 31B IT 62.0 Apr 2026 Official leaderboard
84 gpt-oss-20b 61.9 Aug 2025 Official leaderboard
85 DeepSeek-V3.2-Exp 61.8 Sep 2025 Official leaderboard
86 DeepSeek V4 Flash 0731 61.7 Jul 2026 Official leaderboard
87 Qwen 3.6 Flash 61.5 Apr 2026 Official leaderboard
88 GPT-5.3 Instant 61.1 Mar 2026 Official leaderboard
89 Qwen3.7-Plus 60.7 Jun 2026 Official leaderboard
90 Qwen3.5-27B 60.5 Feb 2026 Official leaderboard
91 Qwen3-Next-80B-A3B 59.6 thinking Sep 2025 Official leaderboard
92 Qwen 3.6 Plus 59.3 Mar 2026 Official leaderboard
93 Qwen3.6 27B 58.9 Apr 2026 Official leaderboard
94 GPT-5 nano 58.8 Aug 2025 Official leaderboard
95 Qwen3 235B-A22B (2507) 58.7 Jul 2025 Official leaderboard
96 MiMo-V2.5-Pro 58.2 Apr 2026 Official leaderboard
97 Qwen3.5-122B-A10B 58.0 Feb 2026 Official leaderboard
98 DeepSeek-V4-Flash 57.3 Apr 2026 Official leaderboard
99 Qwen3-235B-A22B-Thinking Jul 2025 57.1 Jul 2025 Official leaderboard
100 Qwen3-30B-A3B-Thinking Jul 2025 57.0 Jul 2025 Official leaderboard
101 Claude Sonnet 4 57.0 May 2025 Official leaderboard
102 o3-mini 56.5 Jan 2025 Official leaderboard
103 Qwen 3.5 Plus (hosted 397B-A17B) 56.1 Feb 2026 Official leaderboard
104 GPT-4.1 56.0 Apr 2025 Official leaderboard
105 Qwen3.5 397B-A17B 55.6 Feb 2026 Official leaderboard
106 DeepSeek-V3.1 55.5 non-thinking Aug 2025 Official leaderboard
107 Qwen3.5-35B-A3B 55.3 Feb 2026 Official leaderboard
108 DeepSeek-V3.1-Terminus 55.2 Sep 2025 Official leaderboard
109 Nemotron 3.5 Lightning 30B-A3B 54.7 Aug 2026 Official leaderboard
110 Gemma 4 26B A4B 54.3 Apr 2026 Official leaderboard
111 Qwen 3.6 35B-A3B 53.0 Apr 2026 Official leaderboard
112 Qwen 3.5 Flash (hosted 35B-A3B) 52.5 Feb 2026 Official leaderboard
113 Qwen3.5-9B 50.2 Feb 2026 Official leaderboard
114 Llama 4 Maverick 50.2 Apr 2025 Official leaderboard
115 Llama 4 Scout 49.7 Apr 2025 Official leaderboard
116 Mistral Large 3 48.7 Dec 2025 Official leaderboard
117 Qwen3-14B 48.4 Apr 2025 Official leaderboard
118 Qwen3-32B 48.2 Apr 2025 Official leaderboard
119 Qwen3-235B-A22B 46.8 Apr 2025 Official leaderboard
120 Mistral Medium 3.5 45.6 Apr 2026 Official leaderboard
121 Llama 3.3 70B 45.5 Dec 2024 Official leaderboard
122 Qwen3-30B-A3B-Instruct Jul 2025 45.4 Jul 2025 Official leaderboard
123 Gemini 2.5 Flash Jun 2025 45.4 Jun 2025 Official leaderboard
124 Qwen3-30B-A3B 45.0 Apr 2025 Official leaderboard
125 GPT-4o May 2024 44.8 May 2024 Official leaderboard
126 GPT-4.1 mini 43.6 Apr 2025 Official leaderboard
127 DeepSeek-V3 Mar 2025 43.4 Mar 2025 Official leaderboard
128 Llama 3.1-70B 43.2 Jul 2024 Official leaderboard
129 Qwen2.5-72B 42.5 Sep 2024 Official leaderboard
130 GPT-4 Turbo Apr 2024 42.2 Apr 2024 Official leaderboard
131 Qwen3-8B 41.8 Apr 2025 Official leaderboard
132 DeepSeek Chat 41.6 Dec 2024 Official leaderboard
133 Gemma 3n E4B 41.2 Jun 2025 Official leaderboard
134 Gemma 3 27B 40.5 Mar 2025 Official leaderboard
135 Mistral Medium 3.1 39.2 Aug 2025 Official leaderboard
136 Mistral Small 4 39.0 Mar 2026 Official leaderboard
137 Gemma 3 12B 38.3 Mar 2025 Official leaderboard
138 GPT-4 Jun 2023 37.3 Jun 2023 Official leaderboard
139 Cohere Command A 37.3 Mar 2025 Official leaderboard
140 Llama 3.1-8B 33.7 Jul 2024 Official leaderboard
141 Ministral 3 8B Instruct 33.5 Dec 2025 Official leaderboard
142 Gemini 2.5 Flash-Lite Jun 2025 33.4 Jun 2025 Official leaderboard
143 Claude 3 Opus 33.1 Feb 2024 Official leaderboard
144 Ministral 3 14B 32.8 Dec 2025 Official leaderboard
145 GPT-4o mini 32.6 Jul 2024 Official leaderboard
146 Qwen2.5-7B 27.3 Sep 2024 Official leaderboard
147 GPT-4.1 nano 24.3 Apr 2025 Official leaderboard
148 Gemma 3 4B 23.8 Mar 2025 Official leaderboard
149 GPT-3.5 Turbo Nov 2023 20.2 Nov 2023 Official leaderboard
150 Command R+ 18.0 Aug 2024 Official leaderboard
151 Claude 3 Haiku 14.7 Mar 2024 Official leaderboard
152 Ministral 3B 14.6 Oct 2024 Official leaderboard
153 c4ai-command-r-08-2024 13.0 Aug 2024 Official leaderboard
154 Command R7B 9.94 Dec 2024 Official leaderboard

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
Claude Fable 5.2
low high max
89.4
Claude Opus 5.5
low high xhigh
88.4
Claude Sonnet 5.5
low xhigh
85.7
Claude Fable 5
low medium max
85.0
Claude Opus 5
low medium xhigh
83.8
Claude Fable 5.1
low medium high
83.7
GPT-6 Astra
low
81.7
Claude Haiku 5.5
off low medium high xhigh max
81.5
Gemini 3.8 Flash
low
78.2
Gemini 4 Argon
low high
77.3
Gemini 3.7 Flash
low medium
77.0
Claude Sonnet 5
low medium high xhigh max
76.9

Source: Official ACCoRD leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Claude Fable 5.2 xhigh 89.4
Claude Opus 5.5 high 88.4
Claude Fable 5.2 high 88.0
Claude Opus 5.5 max 86.5
Claude Fable 5.2 medium 86.3
Claude Fable 5.2 max 86.2
Claude Opus 5.5 medium 85.7
Claude Sonnet 5.5 xhigh 85.7
Claude Fable 5.2 low 85.5
Claude Sonnet 5.5 max 85.1
Claude Fable 5 max 85.0
Claude Opus 5.5 low 84.9
Claude Opus 5.5 xhigh 84.7
Claude Opus 5 low 83.8
Claude Fable 5.1 medium 83.7
Claude Opus 5 xhigh 83.5
Claude Sonnet 5.5 medium 83.4
Claude Sonnet 5.5 low 82.9
GPT-6 Sol 82.4
Claude Fable 5 medium 82.4
Claude Opus 5 high 82.2
Claude Opus 5 max 82.0
Claude Fable 5 xhigh 81.9
Claude Opus 5 medium 81.7
GPT-6 Astra max 81.7
GPT-6 Astra high 81.6
Claude Haiku 5.5 max 81.5
Claude Sonnet 5.5 high 81.2
GPT-6 Astra low 81.1
Claude Fable 5.1 max 81.0
GPT-6.1 Sol 80.9
Claude Fable 5 high 80.8
GPT-6 Astra xhigh 80.8
Claude Fable 5.1 low 80.8
Claude Fable 5.1 xhigh 80.7
GPT-5.5 80.5
Claude Opus 4.6 80.4
GPT-6 Astra medium 80.4
GPT-5.4 80.0
Claude Fable 5.1 high 79.9
Claude Opus 4.7 79.8
GPT-5.2 79.3
Claude Haiku 5.5 high 79.2
Claude Opus 4.5 79.1
Claude Fable 5 low 78.9
Claude Haiku 5.5 low 78.9
GPT-5.6 Sol Pro 78.9
Muse Spark 1.1 78.4
Gemini 3.8 Flash medium 78.2
Kimi K3.1 Pro 77.5
Gemini 3.8 Flash high 77.5
Claude Opus 4.8 77.4
Muse Spark 1.2 77.4
Gemini 4 Argon low 77.3
Gemini 3.7 Flash high 77.0
Gemini 3.8 Flash low 77.0
Claude Sonnet 5 xhigh 76.9
Claude Haiku 5.5 medium 76.5
Gemini 4 Argon medium 76.2
Muse Wonder 76.2
Gemini 3.7 Flash medium 76.1
GPT-5.6 Sol max 75.9
GPT-5.5 Pro 75.8
Claude Haiku 5.5 xhigh 75.8
Claude Haiku 5.5 off 75.4
Muse Spark 1.3 75.1
GPT-5.6 Sol medium 75.0
Gemini 4 Argon high 75.0
GPT-5.6 Sol xhigh 74.9
GPT-5.6 Sol high 74.7
GPT-5.6 Terra max 74.7
GPT-5.6 Sol low 74.3
Grok 4.7 74.0
o3-pro 73.8
Claude Sonnet 5 max 73.8
Gemini 3.1 Pro medium 73.5
Claude Sonnet 4.6 73.4
Qwen 3.8 27B 73.3
GPT-5 73.1
Claude Sonnet 5 high 73.0
GPT-6 Luna 72.1
Muse Glimmer 72.0
Grok 4.5 71.6
Qwen3.8-2.4T-A95B 71.6
Grok 4.6 71.6
GPT-5.6 Terra medium 71.6
GPT-5 mini 71.5
Qwen 3.8 Max 71.4
o3 71.4
Kimi K2.6 71.4
Inkling 71.2
GPT-5.1 71.2
DeepSeek-V4-Pro 70.9
GLM-5.3 70.9
Qwen 3.6 Max (Preview) 70.8
GPT-5.6 Sol off 70.8
Claude Sonnet 4.5 70.7
Gemini 3.7 Flash low 70.6
GPT-5.4 Mini 70.5
Kimi K3 70.5
GPT-5.6 Terra xhigh 70.4
DeepSeek V4 Pro 0813 70.2
Gemini 3.6 Flash 70.0
Gemini 3.1 Pro low 69.9
GPT-5.6 Terra low 69.8
Grok 4.20 Multi-agent 69.6
Claude Sonnet 5 low 69.5
GPT-5.6 Terra high 69.4
Grok 4.20 68.8
Gemini 3 Flash 68.6
DeepSeek V4.1 Flash 68.1
Gemini 3.5 Flash-Lite 68.0
Claude Sonnet 5 medium 67.6
gpt-oss-120b 67.5
GPT-5.6 Luna high 67.5
GPT-5.6 Luna medium 67.4
GPT-5.4 Nano 67.3
Claude Haiku 4.5 66.7
GPT-5.6 Luna max 66.6
Gemini 3.5 Flash 66.6
Tencent Hy3 66.4
GPT-5.6 Luna xhigh 66.2
Qwen Plus Apr 2025 66.2
GLM-5.2 66.0
GPT-5.2 Instant 66.0
MiniMax-M3 65.9
Claude Opus 4 65.8
Mistral Large 4 65.7
Qwen3.7-Max 65.3
Grok 4.3 65.3
Gemini 3.1 Pro high 65.0
DeepSeek-V3.2 64.6
Nemotron 3 Ultra 64.6
o1 64.2
Gemini 3.1 Flash-Lite 64.1
o4-mini 63.8
Claude Opus 4.1 63.5
GPT-5.6 Luna low 62.9
Qwen3-Max thinking 62.8
Gemini 2.5 Pro Jun 2025 62.7
GPT-5.6 Terra off 62.4
Gemma 4 31B IT 62.0
gpt-oss-20b 61.9
DeepSeek-V3.2-Exp 61.8
DeepSeek V4 Flash 0731 61.7
Qwen 3.6 Flash 61.5
GPT-5.3 Instant 61.1
Qwen3-Max 60.7
Qwen3.7-Plus 60.7
Qwen3.5-27B 60.5
Qwen3-Next-80B-A3B thinking 59.6
Qwen 3.6 Plus 59.3
Qwen3.6 27B 58.9
GPT-5 nano 58.8
MiMo-V2.5-Pro 58.2
Qwen3.5-122B-A10B 58.0
DeepSeek-V4-Flash 57.3
Qwen3-235B-A22B-Thinking Jul 2025 57.1
Qwen3-30B-A3B-Thinking Jul 2025 57.0
Claude Sonnet 4 57.0
o3-mini 56.5
Qwen 3.5 Plus (hosted 397B-A17B) 56.1
GPT-4.1 56.0
Qwen3.5 397B-A17B 55.6
DeepSeek-V3.1 non-thinking 55.5
Qwen3.5-35B-A3B 55.3
DeepSeek-V3.1-Terminus 55.2
Nemotron 3.5 Lightning 30B-A3B 54.7
Gemma 4 26B A4B 54.3
Qwen 3.6 35B-A3B 53.0
Qwen 3.5 Flash (hosted 35B-A3B) 52.5
Qwen3-Next-80B-A3B 52.0
Qwen3.5-9B 50.2
Llama 4 Maverick 50.2
GPT-5.6 Luna off 50.0
Llama 4 Scout 49.7
Mistral Large 3 48.7
Qwen3-14B 48.4
Qwen3-32B 48.2
Qwen3-235B-A22B 46.8
Mistral Medium 3.5 45.6
Llama 3.3 70B 45.5
Qwen3-30B-A3B-Instruct Jul 2025 45.4
Gemini 2.5 Flash Jun 2025 45.4
Qwen3-30B-A3B 45.0
GPT-4o May 2024 44.8
GPT-4.1 mini 43.6
DeepSeek-V3 Mar 2025 43.4
Llama 3.1-70B 43.2
Qwen2.5-72B 42.5
GPT-4 Turbo Apr 2024 42.2
Qwen3-8B 41.8
Gemma 3n E4B 41.2
Gemma 3 27B 40.5
Mistral Medium 3.1 39.2
Mistral Small 4 39.0
Gemma 3 12B 38.3
GPT-4 Jun 2023 37.3
Cohere Command A 37.3
Llama 3.1-8B 33.7
Ministral 3 8B Instruct 33.5
Gemini 2.5 Flash-Lite Jun 2025 33.4
Claude 3 Opus 33.1
Ministral 3 14B 32.8
GPT-4o mini 32.6
Qwen2.5-7B 27.3
GPT-4.1 nano 24.3
Gemma 3 4B 23.8
GPT-3.5 Turbo Nov 2023 20.2
Command R+ 18.0
Claude 3 Haiku 14.7
Ministral 3B 14.6
c4ai-command-r-08-2024 13.0
Command R7B 9.94

3 models have held number one on the official leaderboard since its release in August 2026

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

Anthropic
Own leaderboard 3 leaders
Claude Fable 5.2 89.4

Source: Official ACCoRD leaderboard

Scores over time

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

On the benchmark’s own leaderboard
0.00 20.0 40.0 60.0 80.0 100.0 2023 2024 2025 2026 2027 0.00 20.0 40.0 60.0 80.0 100.0 2023 2024 2025 2026 2027

Source: Official ACCoRD leaderboard

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 Other

All 43 in Other
  • Personal Superintelligence Index (Meta) In use Meta's composite of seven personal and knowledge-work evals: Agents' Last Exam, AutomationBench-AA, FocusBench, JobBench, GDP.pdf, VBT and DRACO.
  • Artificial Analysis Intelligence Index v5 In use Version 5 of Artificial Analysis's index: nine evaluations in four equal categories, most with private test sets, run three times each in its own agent harness. Gemini 4 Carbon 59.0
  • SAGE v1.1 (Vals) In use Models grade handwritten undergraduate math and computer science work against rubrics. Scored by class-balanced accuracy of each rubric check against expert grading. Claude Opus 5.5 91.3%
  • MTEB Fading Massive Text Embedding Benchmark: retrieval, clustering, classification across languages and modalities. sonar 42%
  • Anthropic ECI (AECI) Only in system cards Index that combines a basket of capability benchmarks into one capability score to track trends across Anthropic models. Claude Opus 5.5 169.4
  • SPEED-Bench Only in system cards Measures speculative decoding across varied prompts and domains. Reports the average acceptance length of draft tokens, so higher means faster generation at the same output. Nemotron 3 Ultra 5.15