Benchmark Almanac

OTIS Mock AIME 2024-2025

GPT-5.5 100% leads since April 2026
Not in model cards

Models solve 45 problems from three OTIS Mock AIME exams written by students of the OTIS program. Integer answers are scored by exact match.

Epoch AI added it as a harder successor to MATH Level 5, which was nearing saturation.

Released
December 2024
Kind
Competition problems with answers
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
212
Models scored in Epoch AI’s runs

15 models share the top score of 100%

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

212 models. Scroll sideways for the rest.

  1. 100.0 GPT-5.5
  2. 100.0 GPT-5.5 Pro
  3. 100.0 Claude Fable 5
  4. 100.0 GPT-5.6 Sol
  5. 100.0 Claude Fable 5.1
  6. 100.0 Qwen3.8 Max (0902)
  7. 100.0 GPT-6 Astra
  8. 100.0 Claude Opus 5.5
  9. 100.0 GPT-6 Sol
  10. 100.0 Claude Sonnet 5.5
  11. 100.0 GPT-6.1 Sol
  12. 100.0 Gemini 4 Argon
  13. 100.0 Kimi K3.1 Pro
  14. 100.0 GPT-6.1 Terra
  15. 100.0 Gemini 4 Carbon
  16. 99.7 GPT-5.6 Terra
  17. 99.7 Claude Fable 5.2
  18. 99.4 Qwen 3.8 Max
  19. 99.4 Muse Wonder
  20. 99.4 DeepSeek V4.1 Flash 1017
  21. 99.4 Mistral Large 4
  22. 99.2 Grok 4.6
  23. 99.2 Muse Spark 1.3
  24. 99.2 Doubao Seed Evolving
  25. 98.9 Claude Opus 5
  26. 98.9 Gemini 3.8 Flash
  27. 98.9 GPT-6 Luna
  28. 98.9 Claude Haiku 5.5
  29. 98.6 DeepSeek V4 Pro 0813
  30. 98.3 Claude Opus 4.8
  31. 98.3 GPT-5.6 Luna
  32. 98.3 DeepSeek V4.1 Flash
  33. 98.3 Mistral Large 4 Preview
  34. 98.1 Grok 4.7
  35. 98.1 Muse Spark Open
  36. 97.8 Claude Opus 4.7
  37. 97.8 GPT-5.4
  38. 97.8 Grok 4.5
  39. 97.2 Kimi K3
  40. 97.2 Gemini 3.7 Flash
  41. 96.7 DeepSeek-V4-Pro
  42. 96.1 GPT-5.2
  43. 96.1 Kimi K2.6
  44. 96.1 Doubao Seed 2.1 Turbo
  45. 95.6 Gemini 3.1 Pro
  46. 95.6 Gemini 3 Flash
  47. 95.6 Gemini 3.5 Flash
  48. 95.6 Qwen3.7-Max
  49. 95.6 Kimi K2.7 Code
  50. 95.6 Doubao Seed 2.1 Pro
  51. 94.7 Claude Sonnet 5
  52. 94.4 Claude Opus 4.6
  53. 94.4 DeepSeek V4 Flash 0731
  54. 94.2 Gemini 3.6 Flash
  55. 93.9 Doubao-Seed-2.0 Pro
  56. 93.9 GLM-5.3-Flash
  57. 93.3 Qwen 3.6 Plus
  58. 93.3 GLM-5.1
  59. 93.3 Grok 4.3 Beta
  60. 93.3 Qwen3.7-Plus
  61. 92.2 Grok 4.20
  62. 92.2 Kimi K2.5
  63. 91.4 GPT-5
  64. 91.4 Gemini 3 Pro
  65. 91.1 Qwen 3.6 Max (Preview)
  66. 91.1 Qwen3.6 27B
  67. 91.1 Doubao Seed 2.0 Lite
  68. 91.1 GLM-5.3
  69. 90.0 Inkling-Small
  70. 88.9 Muse Spark
  71. 88.9 gpt-oss-120b
  72. 88.9 Qwen3.5 397B-A17B
  73. 88.9 GPT-5.4 Mini
  74. 88.9 Inkling
  75. 88.6 GPT-5.1
  76. 87.8 DeepSeek-V3.2
  77. 87.8 GPT-5.4 Nano
  78. 86.7 Qwen3-235B-A22B-Thinking
  79. 86.7 GPT-5 mini
  80. 86.7 Qwen 3.5 Plus (hosted 397B-A17B)
  81. 86.7 Qwen 3.6 35B-A3B
  82. 86.7 Nemotron 3 Ultra
  83. 86.7 Qwen3.7 Flash
  84. 86.4 GLM-5.2
  85. 86.1 Claude Opus 4.5
  86. 85.8 Claude Sonnet 4.6
  87. 84.7 Gemini 2.5 Pro
  88. 84.4 o3
  89. 84.4 Qwen 3.5 Flash (hosted 35B-A3B)
  90. 84.4 Qwen 3.6 Flash
  91. 84.0 Grok 4
  92. 83.3 GLM-4.7
  93. 83.1 Kimi K2 Thinking
  94. 82.2 Gemma 4 26B A4B
  95. 81.7 o4-mini
  96. 81.1 GPT-5 nano
  97. 80.0 GLM-5
  98. 80.0 Gemini 3.1 Flash-Lite
  99. 77.8 Grok-3 mini
  100. 77.8 Claude Sonnet 4.5
  101. 76.9 o3-mini
  102. 73.3 o1
  103. 73.3 Qwen3-Max
  104. 73.3 Gemma 4 31B IT
  105. 73.1 Gemini 2.5 Flash
  106. 71.1 Claude Sonnet 4
  107. 71.1 MiniMax-M3
  108. 71.1 Gemini 3.5 Flash-Lite
  109. 70.8 Gemini 2.5 Flash
  110. 70.3 Qwen3-30B-A3B-Thinking
  111. 70.0 Qwen3.5-35B-A3B
  112. 69.4 Doubao-Seed-1.6
  113. 68.9 Claude Opus 4.1
  114. 68.1 GPT-5.5 Instant
  115. 67.5 seed-oss-36b-instruct
  116. 66.9 Qwen3-32B
  117. 66.7 Claude Haiku 4.5
  118. 66.4 Qwen3-14B
  119. 66.4 DeepSeek-R1
  120. 65.3 gpt-oss-20b
  121. 64.4 Claude Opus 4
  122. 62.8 Qwen3-30B-A3B
  123. 62.2 Qwen3-30B-A3B-Instruct
  124. 61.7 Qwen3.5-9B
  125. 59.2 QwQ-32B
  126. 58.3 glm-4.7-flash
  127. 57.8 Gemini 2.0 Flash Thinking
  128. 57.8 Claude 3.7 Sonnet
  129. 56.1 Qwen3-8B
  130. 55.8 Qwen3.5-4B
  131. 55.6 DeepSeek-R1-Distill-Qwen-32B
  132. 55.6 Grok 3
  133. 53.3 DeepSeek-R1
  134. 52.2 qwen3-4b-instruct-2507
  135. 51.4 DeepSeek-R1-Distill-Llama-70B
  136. 50.6 DeepSeek-R1-Distill-Qwen-14B
  137. 46.9 o1-mini
  138. 44.7 GPT-4.1 mini
  139. 43.9 deepseek-r1-0528-qwen3-8b
  140. 38.3 GPT-4.1
  141. 37.8 GPT-4.5
  142. 37.8 DeepSeek-V3
  143. 32.2 Mistral Medium 3
  144. 31.1 o1-preview
  145. 31.1 Gemini 2.0 Flash
  146. 30.3 Mistral Small 3.2
  147. 30.0 Magistral Small 1.0
  148. 28.9 GPT-4.1 nano
  149. 28.1 Magistral Small 1.2
  150. 25.0 glm-4.7-flash_none
  151. 23.1 Gemini 1.5 Pro
  152. 22.5 Gemma 3 27B
  153. 21.4 DeepSeek-R1-Distill-Qwen-1.5B
  154. 20.6 Llama 4 Maverick
  155. 17.8 Qwen Plus
  156. 16.7 Gemma 3 12B
  157. 16.3 Gemini 1.5 Flash
  158. 16.1 Qwen2.5-Max
  159. 15.8 DeepSeek-V3
  160. 13.8 Phi-4
  161. 11.5 Grok-2
  162. 9.72 Llama 3.1-405B
  163. 8.47 Mistral Large 2
  164. 8.47 Claude 3.5 Sonnet (October 2024)
  165. 8.06 Qwen2.5-72B
  166. 8.06 Qwen3-1.7B
  167. 7.78 Mistral Large 2
  168. 7.78 Llama 4 Scout
  169. 7.50 Gemma 3 4B
  170. 7.36 Qwen2.5-32B
  171. 6.94 GPT-4o mini
  172. 6.81 Gemini 1.5 Pro
  173. 6.67 GPT-4 Turbo
  174. 6.67 Mistral Small 3
  175. 6.53 Claude 3.5 Sonnet
  176. 6.39 GPT-4o
  177. 6.25 GPT-4o
  178. 6.25 GPT-4o
  179. 6.11 Qwen-Turbo
  180. 5.83 Mistral Small 3.1
  181. 5.14 Llama 3.3 70B
  182. 4.72 Claude 3 Opus
  183. 4.58 Gemini 1.5 Flash 8B
  184. 4.44 Tulu 3 (Tülu 3) 70B
  185. 4.31 Llama 3-70B
  186. 4.31 Claude 3.5 Haiku
  187. 3.89 Gemini 1.5 Flash
  188. 3.61 Llama 3.1-70B
  189. 2.78 granite-4.0-micro
  190. 2.64 Llama 3.2 90B
  191. 2.50 Claude 2
  192. 2.50 Claude 3 Sonnet
  193. 2.50 Hermes 2 Theta Llama-3 70B
  194. 2.50 Qwen2.5-7B
  195. 2.22 GPT-3.5 Turbo
  196. 1.94 Claude 2.1
  197. 1.94 Mistral Large
  198. 1.94 Llama 3-8B
  199. 1.81 Claude 3 Haiku
  200. 1.67 Llama 3.1-8B
  201. 1.39 Gemma 2 27B
  202. 1.11 GPT-4
  203. 1.11 Gemini 1.0 Pro
  204. 1.11 Gemma 3 1B
  205. 0.83 DeepSeek LLM 67B
  206. 0.83 granite-4.0-1b
  207. 0.56 GPT-4
  208. 0.56 Gemma 2 9B
  209. 0.56 Llama 3.2 1B
  210. 0.28 Mistral 7B v0.3
  211. 0.28 granite-4.0-350m
  212. 0.0 Llama 2-70B
  1. 1 GPT-5.5 xhigh 100%
  2. 2 GPT-5.5 Pro xhigh 100%
  3. 3 Claude Fable 5 high 100%
  4. 4 GPT-5.6 Sol max 100%
  5. 5 Claude Fable 5.1 max 100%
  6. 6 Qwen3.8 Max (0902) xhigh 100%
  7. 7 GPT-6 Astra max 100%
  8. 8 Claude Opus 5.5 max 100%
  9. 9 GPT-6 Sol max 100%
  10. 10 Claude Sonnet 5.5 max 100%
  11. 11 GPT-6.1 Sol max 100%
  12. 12 Gemini 4 Argon high 100%
  13. 13 Kimi K3.1 Pro Together, max 100%
  14. 14 GPT-6.1 Terra max 100%
  15. 15 Gemini 4 Carbon high 100%
  16. 16 GPT-5.6 Terra max 99.7%
  17. 17 Claude Fable 5.2 max 99.7%
  18. 18 Qwen 3.8 Max xhigh 99.4%
  19. 19 Muse Wonder max 99.4%
  20. 20 DeepSeek V4.1 Flash 1017 max 99.4%
  21. 21 Mistral Large 4 high 99.4%
  22. 22 Grok 4.6 xhigh 99.2%
  23. 23 Muse Spark 1.3 xhigh 99.2%
  24. 24 Doubao Seed Evolving Nov 2026, high 99.2%
  25. 25 Claude Opus 5 max 98.9%
  26. 26 Gemini 3.8 Flash high 98.9%
  27. 27 GPT-6 Luna max 98.9%
  28. 28 Claude Haiku 5.5 xhigh 98.9%
  29. 29 DeepSeek V4 Pro 0813 max 98.6%
  30. 30 Claude Opus 4.8 max 98.3%
  31. 31 GPT-5.6 Luna max 98.3%
  32. 32 DeepSeek V4.1 Flash max 98.3%
  33. 33 Mistral Large 4 Preview high 98.3%
  34. 34 Grok 4.7 xhigh 98.1%
  35. 35 Muse Spark Open xhigh 98.1%
  36. 36 Claude Opus 4.7 xhigh 97.8%
  37. 37 GPT-5.4 high 97.8%
  38. 38 Grok 4.5 high 97.8%
  39. 39 Kimi K3 max 97.2%
  40. 40 Gemini 3.7 Flash high 97.2%
  41. 41 DeepSeek-V4-Pro max 96.7%
  42. 42 GPT-5.2 high 96.1%
  43. 43 Kimi K2.6 96.1%
  44. 44 Doubao Seed 2.1 Turbo high 96.1%
  45. 45 Gemini 3.1 Pro 95.6%
  46. 46 Gemini 3 Flash high 95.6%
  47. 47 Gemini 3.5 Flash high 95.6%
  48. 48 Qwen3.7-Max 95.6%
  49. 49 Kimi K2.7 Code 95.6%
  50. 50 Doubao Seed 2.1 Pro high 95.6%
  51. 51 Claude Sonnet 5 xhigh 94.7%
  52. 52 Claude Opus 4.6 64k thinking 94.4%
  53. 53 DeepSeek V4 Flash 0731 max 94.4%
  54. 54 Gemini 3.6 Flash high 94.2%
  55. 55 Doubao-Seed-2.0 Pro high 93.9%
  56. 56 GLM-5.3-Flash max 93.9%
  57. 57 Qwen 3.6 Plus 2026-04-02 93.3%
  58. 58 GLM-5.1 93.3%
  59. 59 Grok 4.3 Beta high 93.3%
  60. 60 Qwen3.7-Plus 93.3%
  61. 61 Grok 4.20 92.2%
  62. 62 Kimi K2.5 Fireworks 92.2%
  63. 63 GPT-5 high 91.4%
  64. 64 Gemini 3 Pro 91.4%
  65. 65 Qwen 3.6 Max (Preview) 91.1%
  66. 66 Qwen3.6 27B 91.1%
  67. 67 Doubao Seed 2.0 Lite Apr 2026, high 91.1%
  68. 68 GLM-5.3 max 91.1%
  69. 69 Inkling-Small xhigh 90%
  70. 70 Muse Spark 88.9%
  71. 71 gpt-oss-120b high 88.9%
  72. 72 Qwen3.5 397B-A17B 88.9%
  73. 73 GPT-5.4 Mini xhigh 88.9%
  74. 74 Inkling xhigh 88.9%
  75. 75 GPT-5.1 high 88.6%
  76. 76 DeepSeek-V3.2 Thinking 87.8%
  77. 77 GPT-5.4 Nano high 87.8%
  78. 78 Qwen3-235B-A22B-Thinking Jul 2025 86.7%
  79. 79 GPT-5 mini high 86.7%
  80. 80 Qwen 3.5 Plus (hosted 397B-A17B) 86.7%
  81. 81 Qwen 3.6 35B-A3B 86.7%
  82. 82 Nemotron 3 Ultra 86.7%
  83. 83 Qwen3.7 Flash 86.7%
  84. 84 GLM-5.2 max 86.4%
  85. 85 Claude Opus 4.5 32k thinking 86.1%
  86. 86 Claude Sonnet 4.6 32k thinking 85.8%
  87. 87 Gemini 2.5 Pro Jun 2025 84.7%
  88. 88 o3 medium 84.4%
  89. 89 Qwen 3.5 Flash (hosted 35B-A3B) 84.4%
  90. 90 Qwen 3.6 Flash 2026-04-16 84.4%
  91. 91 Grok 4 84%
  92. 92 GLM-4.7 83.3%
  93. 93 Kimi K2 Thinking 83.1%
  94. 94 Gemma 4 26B A4B minimal 82.2%
  95. 95 o4-mini high 81.7%
  96. 96 GPT-5 nano high 81.1%
  97. 97 GLM-5 80%
  98. 98 Gemini 3.1 Flash-Lite high 80%
  99. 99 Grok-3 mini high 77.8%
  100. 100 Claude Sonnet 4.5 59k thinking 77.8%
  101. 101 o3-mini high 76.9%
  102. 102 o1 high 73.3%
  103. 103 Qwen3-Max 73.3%
  104. 104 Gemma 4 31B IT minimal 73.3%
  105. 105 Gemini 2.5 Flash Apr 2025 73.1%
  106. 106 Claude Sonnet 4 32K 71.1%
  107. 107 MiniMax-M3 71.1%
  108. 108 Gemini 3.5 Flash-Lite high 71.1%
  109. 109 Gemini 2.5 Flash May 2025 70.8%
  110. 110 Qwen3-30B-A3B-Thinking Jul 2025 70.3%
  111. 111 Qwen3.5-35B-A3B none 70%
  112. 112 Doubao-Seed-1.6 thinking 69.4%
  113. 113 Claude Opus 4.1 27K 68.9%
  114. 114 GPT-5.5 Instant 68.1%
  115. 115 seed-oss-36b-instruct 67.5%
  116. 116 Qwen3-32B 66.9%
  117. 117 Claude Haiku 4.5 32K 66.7%
  118. 118 Qwen3-14B 66.4%
  119. 119 DeepSeek-R1 May 2025 66.4%
  120. 120 gpt-oss-20b medium 65.3%
  121. 121 Claude Opus 4 27K 64.4%
  122. 122 Qwen3-30B-A3B 62.8%
  123. 123 Qwen3-30B-A3B-Instruct Jul 2025 62.2%
  124. 124 Qwen3.5-9B none 61.7%
  125. 125 QwQ-32B 59.2%
  126. 126 glm-4.7-flash 58.3%
  127. 127 Gemini 2.0 Flash Thinking Jan 2025 57.8%
  128. 128 Claude 3.7 Sonnet 64k thinking 57.8%
  129. 129 Qwen3-8B 56.1%
  130. 130 Qwen3.5-4B none 55.8%
  131. 131 DeepSeek-R1-Distill-Qwen-32B 55.6%
  132. 132 Grok 3 55.6%
  133. 133 DeepSeek-R1 53.3%
  134. 134 qwen3-4b-instruct-2507 52.2%
  135. 135 DeepSeek-R1-Distill-Llama-70B 51.4%
  136. 136 DeepSeek-R1-Distill-Qwen-14B 50.6%
  137. 137 o1-mini high 46.9%
  138. 138 GPT-4.1 mini 44.7%
  139. 139 deepseek-r1-0528-qwen3-8b 43.9%
  140. 140 GPT-4.1 38.3%
  141. 141 GPT-4.5 37.8%
  142. 142 DeepSeek-V3 Mar 2025 37.8%
  143. 143 Mistral Medium 3 32.2%
  144. 144 o1-preview 31.1%
  145. 145 Gemini 2.0 Flash Feb 2025 31.1%
  146. 146 Mistral Small 3.2 30.3%
  147. 147 Magistral Small 1.0 30%
  148. 148 GPT-4.1 nano 28.9%
  149. 149 Magistral Small 1.2 28.1%
  150. 150 glm-4.7-flash_none none 25%
  151. 151 Gemini 1.5 Pro Sep 2024 23.1%
  152. 152 Gemma 3 27B 22.5%
  153. 153 DeepSeek-R1-Distill-Qwen-1.5B 21.4%
  154. 154 Llama 4 Maverick FP8 20.6%
  155. 155 Qwen Plus Jan 2025 17.8%
  156. 156 Gemma 3 12B 16.7%
  157. 157 Gemini 1.5 Flash Sep 2024 16.3%
  158. 158 Qwen2.5-Max 16.1%
  159. 159 DeepSeek-V3 15.8%
  160. 160 Phi-4 13.8%
  161. 161 Grok-2 Dec 2024 11.5%
  162. 162 Llama 3.1-405B 9.72%
  163. 163 Mistral Large 2 Jul 2024 8.47%
  164. 164 Claude 3.5 Sonnet (October 2024) 8.47%
  165. 165 Qwen2.5-72B 8.06%
  166. 166 Qwen3-1.7B none 8.06%
  167. 167 Mistral Large 2 Nov 2024 7.78%
  168. 168 Llama 4 Scout 7.78%
  169. 169 Gemma 3 4B 7.50%
  170. 170 Qwen2.5-32B 7.36%
  171. 171 GPT-4o mini 6.94%
  172. 172 Gemini 1.5 Pro May 2024 6.81%
  173. 173 GPT-4 Turbo Apr 2024 6.67%
  174. 174 Mistral Small 3 6.67%
  175. 175 Claude 3.5 Sonnet Jun 2024 6.53%
  176. 176 GPT-4o Aug 2024 6.39%
  177. 177 GPT-4o May 2024 6.25%
  178. 178 GPT-4o Nov 2024 6.25%
  179. 179 Qwen-Turbo 6.11%
  180. 180 Mistral Small 3.1 5.83%
  181. 181 Llama 3.3 70B 5.14%
  182. 182 Claude 3 Opus 4.72%
  183. 183 Gemini 1.5 Flash 8B 4.58%
  184. 184 Tulu 3 (Tülu 3) 70B 4.44%
  185. 185 Llama 3-70B 4.31%
  186. 186 Claude 3.5 Haiku Oct 2024 4.31%
  187. 187 Gemini 1.5 Flash May 2024 3.89%
  188. 188 Llama 3.1-70B 3.61%
  189. 189 granite-4.0-micro 2.78%
  190. 190 Llama 3.2 90B 2.64%
  191. 191 Claude 2 2.50%
  192. 192 Claude 3 Sonnet 2.50%
  193. 193 Hermes 2 Theta Llama-3 70B 2.50%
  194. 194 Qwen2.5-7B 2.50%
  195. 195 GPT-3.5 Turbo Jan 2024 2.22%
  196. 196 Claude 2.1 1.94%
  197. 197 Mistral Large 1.94%
  198. 198 Llama 3-8B 1.94%
  199. 199 Claude 3 Haiku 1.81%
  200. 200 Llama 3.1-8B 1.67%
  201. 201 Gemma 2 27B 1.39%
  202. 202 GPT-4 Jun 2023 1.11%
  203. 203 Gemini 1.0 Pro 1.11%
  204. 204 Gemma 3 1B 1.11%
  205. 205 DeepSeek LLM 67B 0.83%
  206. 206 granite-4.0-1b 0.83%
  207. 207 GPT-4 Mar 2023 0.56%
  208. 208 Gemma 2 9B 0.56%
  209. 209 Llama 3.2 1B 0.56%
  210. 210 Mistral 7B v0.3 0.28%
  211. 211 granite-4.0-350m 0.28%
  212. 212 Llama 2-70B 0%

Source: Epoch AI runs (CC BY 4.0)

#ModelScoreSettingDateSource
1 GPT-5.5 100% xhigh Apr 2026 Epoch AI run
2 GPT-5.5 Pro 100% xhigh Apr 2026 Epoch AI run
3 Claude Fable 5 100% high Jun 2026 Epoch AI run
4 GPT-5.6 Sol 100% max Jul 2026 Epoch AI run
5 Claude Fable 5.1 100% max Sep 2026 Epoch AI run
6 Qwen3.8 Max (0902) 100% xhigh Sep 2026 Epoch AI run
7 GPT-6 Astra 100% max Sep 2026 Epoch AI run
8 Claude Opus 5.5 100% max Sep 2026 Epoch AI run
9 GPT-6 Sol 100% max Sep 2026 Epoch AI run
10 Claude Sonnet 5.5 100% max Sep 2026 Epoch AI run
11 GPT-6.1 Sol 100% max Sep 2026 Epoch AI run
12 Gemini 4 Argon 100% high Sep 2026 Epoch AI run
13 Kimi K3.1 Pro 100% Together, max Oct 2026 Epoch AI run
14 GPT-6.1 Terra 100% max Oct 2026 Epoch AI run
15 Gemini 4 Carbon 100% high Oct 2026 Epoch AI run
16 GPT-5.6 Terra 99.7% max Jul 2026 Epoch AI run
17 Claude Fable 5.2 99.7% max Nov 2026 Epoch AI run
18 Qwen 3.8 Max 99.4% xhigh Aug 2026 Epoch AI run
19 Muse Wonder 99.4% max Oct 2026 Epoch AI run
20 DeepSeek V4.1 Flash 1017 99.4% max Oct 2026 Epoch AI run
21 Mistral Large 4 99.4% high Oct 2026 Epoch AI run
22 Grok 4.6 99.2% xhigh Aug 2026 Epoch AI run
23 Muse Spark 1.3 99.2% xhigh Sep 2026 Epoch AI run
24 Doubao Seed Evolving Nov 2026 99.2% high Nov 2026 Epoch AI run
25 Claude Opus 5 98.9% max Jul 2026 Epoch AI run
26 Gemini 3.8 Flash 98.9% high Sep 2026 Epoch AI run
27 GPT-6 Luna 98.9% max Sep 2026 Epoch AI run
28 Claude Haiku 5.5 98.9% xhigh Oct 2026 Epoch AI run
29 DeepSeek V4 Pro 0813 98.6% max Aug 2026 Epoch AI run
30 Claude Opus 4.8 98.3% max May 2026 Epoch AI run
31 GPT-5.6 Luna 98.3% max Jul 2026 Epoch AI run
32 DeepSeek V4.1 Flash 98.3% max Sep 2026 Epoch AI run
33 Mistral Large 4 Preview 98.3% high Oct 2026 Epoch AI run
34 Grok 4.7 98.1% xhigh Sep 2026 Epoch AI run
35 Muse Spark Open 98.1% xhigh Oct 2026 Epoch AI run
36 Claude Opus 4.7 97.8% xhigh Apr 2026 Epoch AI run
37 GPT-5.4 97.8% high Mar 2026 Epoch AI run
38 Grok 4.5 97.8% high Jul 2026 Epoch AI run
39 Kimi K3 97.2% max Jul 2026 Epoch AI run
40 Gemini 3.7 Flash 97.2% high Aug 2026 Epoch AI run
41 DeepSeek-V4-Pro 96.7% max Apr 2026 Epoch AI run
42 GPT-5.2 96.1% high Dec 2025 Epoch AI run
43 Kimi K2.6 96.1% Apr 2026 Epoch AI run
44 Doubao Seed 2.1 Turbo 96.1% high Jun 2026 Epoch AI run
45 Gemini 3.1 Pro 95.6% Feb 2026 Epoch AI run
46 Gemini 3 Flash 95.6% high Dec 2025 Epoch AI run
47 Gemini 3.5 Flash 95.6% high May 2026 Epoch AI run
48 Qwen3.7-Max 95.6% May 2026 Epoch AI run
49 Kimi K2.7 Code 95.6% Jun 2026 Epoch AI run
50 Doubao Seed 2.1 Pro 95.6% high Jun 2026 Epoch AI run
51 Claude Sonnet 5 94.7% xhigh Jun 2026 Epoch AI run
52 Claude Opus 4.6 94.4% 64k thinking Feb 2026 Epoch AI run
53 DeepSeek V4 Flash 0731 94.4% max Jul 2026 Epoch AI run
54 Gemini 3.6 Flash 94.2% high Jul 2026 Epoch AI run
55 Doubao-Seed-2.0 Pro 93.9% high Feb 2026 Epoch AI run
56 GLM-5.3-Flash 93.9% max Aug 2026 Epoch AI run
57 Qwen 3.6 Plus 93.3% 2026-04-02 Mar 2026 Epoch AI run
58 GLM-5.1 93.3% Apr 2026 Epoch AI run
59 Grok 4.3 Beta 93.3% high Apr 2026 Epoch AI run
60 Qwen3.7-Plus 93.3% Jun 2026 Epoch AI run
61 Grok 4.20 92.2% Feb 2026 Epoch AI run
62 Kimi K2.5 92.2% Fireworks Jan 2026 Epoch AI run
63 GPT-5 91.4% high Aug 2025 Epoch AI run
64 Gemini 3 Pro 91.4% Nov 2025 Epoch AI run
65 Qwen 3.6 Max (Preview) 91.1% Apr 2026 Epoch AI run
66 Qwen3.6 27B 91.1% Apr 2026 Epoch AI run
67 Doubao Seed 2.0 Lite Apr 2026 91.1% high Apr 2026 Epoch AI run
68 GLM-5.3 91.1% max Aug 2026 Epoch AI run
69 Inkling-Small 90% xhigh Jul 2026 Epoch AI run
70 Muse Spark 88.9% Apr 2026 Epoch AI run
71 gpt-oss-120b 88.9% high Aug 2025 Epoch AI run
72 Qwen3.5 397B-A17B 88.9% Feb 2026 Epoch AI run
73 GPT-5.4 Mini 88.9% xhigh Mar 2026 Epoch AI run
74 Inkling 88.9% xhigh Jul 2026 Epoch AI run
75 GPT-5.1 88.6% high Nov 2025 Epoch AI run
76 DeepSeek-V3.2 87.8% Thinking Dec 2025 Epoch AI run
77 GPT-5.4 Nano 87.8% high Mar 2026 Epoch AI run
78 Qwen3-235B-A22B-Thinking Jul 2025 86.7% Jul 2025 Epoch AI run
79 GPT-5 mini 86.7% high Aug 2025 Epoch AI run
80 Qwen 3.5 Plus (hosted 397B-A17B) 86.7% Feb 2026 Epoch AI run
81 Qwen 3.6 35B-A3B 86.7% Apr 2026 Epoch AI run
82 Nemotron 3 Ultra 86.7% Jun 2026 Epoch AI run
83 Qwen3.7 Flash 86.7% Jul 2026 Epoch AI run
84 GLM-5.2 86.4% max Jun 2026 Epoch AI run
85 Claude Opus 4.5 86.1% 32k thinking Nov 2025 Epoch AI run
86 Claude Sonnet 4.6 85.8% 32k thinking Feb 2026 Epoch AI run
87 Gemini 2.5 Pro Jun 2025 84.7% Jun 2025 Epoch AI run
88 o3 84.4% medium Apr 2025 Epoch AI run
89 Qwen 3.5 Flash (hosted 35B-A3B) 84.4% Feb 2026 Epoch AI run
90 Qwen 3.6 Flash 84.4% 2026-04-16 Apr 2026 Epoch AI run
91 Grok 4 84% Jul 2025 Epoch AI run
92 GLM-4.7 83.3% Dec 2025 Epoch AI run
93 Kimi K2 Thinking 83.1% Nov 2025 Epoch AI run
94 Gemma 4 26B A4B 82.2% minimal Apr 2026 Epoch AI run
95 o4-mini 81.7% high Apr 2025 Epoch AI run
96 GPT-5 nano 81.1% high Aug 2025 Epoch AI run
97 GLM-5 80% Feb 2026 Epoch AI run
98 Gemini 3.1 Flash-Lite 80% high Mar 2026 Epoch AI run
99 Grok-3 mini 77.8% high Apr 2025 Epoch AI run
100 Claude Sonnet 4.5 77.8% 59k thinking Sep 2025 Epoch AI run
101 o3-mini 76.9% high Jan 2025 Epoch AI run
102 o1 73.3% high Dec 2024 Epoch AI run
103 Qwen3-Max 73.3% Sep 2025 Epoch AI run
104 Gemma 4 31B IT 73.3% minimal Apr 2026 Epoch AI run
105 Gemini 2.5 Flash Apr 2025 73.1% Apr 2025 Epoch AI run
106 Claude Sonnet 4 71.1% 32K May 2025 Epoch AI run
107 MiniMax-M3 71.1% Jun 2026 Epoch AI run
108 Gemini 3.5 Flash-Lite 71.1% high Jul 2026 Epoch AI run
109 Gemini 2.5 Flash May 2025 70.8% May 2025 Epoch AI run
110 Qwen3-30B-A3B-Thinking Jul 2025 70.3% Jul 2025 Epoch AI run
111 Qwen3.5-35B-A3B 70% none Feb 2026 Epoch AI run
112 Doubao-Seed-1.6 69.4% thinking Jun 2025 Epoch AI run
113 Claude Opus 4.1 68.9% 27K Aug 2025 Epoch AI run
114 GPT-5.5 Instant 68.1% May 2026 Epoch AI run
115 seed-oss-36b-instruct 67.5% Aug 2025 Epoch AI run
116 Qwen3-32B 66.9% Apr 2025 Epoch AI run
117 Claude Haiku 4.5 66.7% 32K Oct 2025 Epoch AI run
118 Qwen3-14B 66.4% Apr 2025 Epoch AI run
119 DeepSeek-R1 May 2025 66.4% May 2025 Epoch AI run
120 gpt-oss-20b 65.3% medium Aug 2025 Epoch AI run
121 Claude Opus 4 64.4% 27K May 2025 Epoch AI run
122 Qwen3-30B-A3B 62.8% Apr 2025 Epoch AI run
123 Qwen3-30B-A3B-Instruct Jul 2025 62.2% Jul 2025 Epoch AI run
124 Qwen3.5-9B 61.7% none Feb 2026 Epoch AI run
125 QwQ-32B 59.2% Mar 2025 Epoch AI run
126 glm-4.7-flash 58.3% Jan 2026 Epoch AI run
127 Gemini 2.0 Flash Thinking Jan 2025 57.8% Jan 2025 Epoch AI run
128 Claude 3.7 Sonnet 57.8% 64k thinking Feb 2025 Epoch AI run
129 Qwen3-8B 56.1% Apr 2025 Epoch AI run
130 Qwen3.5-4B 55.8% none Feb 2026 Epoch AI run
131 DeepSeek-R1-Distill-Qwen-32B 55.6% Jan 2025 Epoch AI run
132 Grok 3 55.6% Apr 2025 Epoch AI run
133 DeepSeek-R1 53.3% Jan 2025 Epoch AI run
134 qwen3-4b-instruct-2507 52.2% Aug 2025 Epoch AI run
135 DeepSeek-R1-Distill-Llama-70B 51.4% Jan 2025 Epoch AI run
136 DeepSeek-R1-Distill-Qwen-14B 50.6% Jan 2025 Epoch AI run
137 o1-mini 46.9% high Sep 2024 Epoch AI run
138 GPT-4.1 mini 44.7% Apr 2025 Epoch AI run
139 deepseek-r1-0528-qwen3-8b 43.9% May 2025 Epoch AI run
140 GPT-4.1 38.3% Apr 2025 Epoch AI run
141 GPT-4.5 37.8% Feb 2025 Epoch AI run
142 DeepSeek-V3 Mar 2025 37.8% Mar 2025 Epoch AI run
143 Mistral Medium 3 32.2% May 2025 Epoch AI run
144 o1-preview 31.1% Sep 2024 Epoch AI run
145 Gemini 2.0 Flash Feb 2025 31.1% Feb 2025 Epoch AI run
146 Mistral Small 3.2 30.3% Jun 2025 Epoch AI run
147 Magistral Small 1.0 30% Jun 2025 Epoch AI run
148 GPT-4.1 nano 28.9% Apr 2025 Epoch AI run
149 Magistral Small 1.2 28.1% Sep 2025 Epoch AI run
150 glm-4.7-flash_none 25% none Epoch AI run
151 Gemini 1.5 Pro Sep 2024 23.1% Sep 2024 Epoch AI run
152 Gemma 3 27B 22.5% Mar 2025 Epoch AI run
153 DeepSeek-R1-Distill-Qwen-1.5B 21.4% Jan 2025 Epoch AI run
154 Llama 4 Maverick 20.6% FP8 Apr 2025 Epoch AI run
155 Qwen Plus Jan 2025 17.8% Jan 2025 Epoch AI run
156 Gemma 3 12B 16.7% Mar 2025 Epoch AI run
157 Gemini 1.5 Flash Sep 2024 16.3% Sep 2024 Epoch AI run
158 Qwen2.5-Max 16.1% Jan 2025 Epoch AI run
159 DeepSeek-V3 15.8% Dec 2024 Epoch AI run
160 Phi-4 13.8% Dec 2024 Epoch AI run
161 Grok-2 Dec 2024 11.5% Dec 2024 Epoch AI run
162 Llama 3.1-405B 9.72% Jul 2024 Epoch AI run
163 Mistral Large 2 Jul 2024 8.47% Jul 2024 Epoch AI run
164 Claude 3.5 Sonnet (October 2024) 8.47% Oct 2024 Epoch AI run
165 Qwen2.5-72B 8.06% Sep 2024 Epoch AI run
166 Qwen3-1.7B 8.06% none Apr 2025 Epoch AI run
167 Mistral Large 2 Nov 2024 7.78% Nov 2024 Epoch AI run
168 Llama 4 Scout 7.78% Apr 2025 Epoch AI run
169 Gemma 3 4B 7.50% Mar 2025 Epoch AI run
170 Qwen2.5-32B 7.36% Sep 2024 Epoch AI run
171 GPT-4o mini 6.94% Jul 2024 Epoch AI run
172 Gemini 1.5 Pro May 2024 6.81% May 2024 Epoch AI run
173 GPT-4 Turbo Apr 2024 6.67% Apr 2024 Epoch AI run
174 Mistral Small 3 6.67% Jan 2025 Epoch AI run
175 Claude 3.5 Sonnet 6.53% Jun 2024 Jun 2024 Epoch AI run
176 GPT-4o Aug 2024 6.39% Aug 2024 Epoch AI run
177 GPT-4o May 2024 6.25% May 2024 Epoch AI run
178 GPT-4o Nov 2024 6.25% Nov 2024 Epoch AI run
179 Qwen-Turbo 6.11% Nov 2024 Epoch AI run
180 Mistral Small 3.1 5.83% Mar 2025 Epoch AI run
181 Llama 3.3 70B 5.14% Dec 2024 Epoch AI run
182 Claude 3 Opus 4.72% Feb 2024 Epoch AI run
183 Gemini 1.5 Flash 8B 4.58% Oct 2024 Epoch AI run
184 Tulu 3 (Tülu 3) 70B 4.44% Nov 2024 Epoch AI run
185 Llama 3-70B 4.31% Apr 2024 Epoch AI run
186 Claude 3.5 Haiku 4.31% Oct 2024 Oct 2024 Epoch AI run
187 Gemini 1.5 Flash May 2024 3.89% May 2024 Epoch AI run
188 Llama 3.1-70B 3.61% Jul 2024 Epoch AI run
189 granite-4.0-micro 2.78% Oct 2025 Epoch AI run
190 Llama 3.2 90B 2.64% Sep 2024 Epoch AI run
191 Claude 2 2.50% Jul 2023 Epoch AI run
192 Claude 3 Sonnet 2.50% Feb 2024 Epoch AI run
193 Hermes 2 Theta Llama-3 70B 2.50% Jun 2024 Epoch AI run
194 Qwen2.5-7B 2.50% Sep 2024 Epoch AI run
195 GPT-3.5 Turbo Jan 2024 2.22% Jan 2024 Epoch AI run
196 Claude 2.1 1.94% Nov 2023 Epoch AI run
197 Mistral Large 1.94% Feb 2024 Epoch AI run
198 Llama 3-8B 1.94% Apr 2024 Epoch AI run
199 Claude 3 Haiku 1.81% Mar 2024 Epoch AI run
200 Llama 3.1-8B 1.67% Jul 2024 Epoch AI run
201 Gemma 2 27B 1.39% Jun 2024 Epoch AI run
202 GPT-4 Jun 2023 1.11% Jun 2023 Epoch AI run
203 Gemini 1.0 Pro 1.11% Dec 2023 Epoch AI run
204 Gemma 3 1B 1.11% Mar 2025 Epoch AI run
205 DeepSeek LLM 67B 0.83% Nov 2023 Epoch AI run
206 granite-4.0-1b 0.83% Oct 2025 Epoch AI run
207 GPT-4 Mar 2023 0.56% Mar 2023 Epoch AI run
208 Gemma 2 9B 0.56% Jun 2024 Epoch AI run
209 Llama 3.2 1B 0.56% Sep 2024 Epoch AI run
210 Mistral 7B v0.3 0.28% May 2024 Epoch AI run
211 granite-4.0-350m 0.28% Oct 2025 Epoch AI run
212 Llama 2-70B 0% Jul 2023 Epoch AI run

13 models have held number one in Epoch AI’s runs since its release in December 2024

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

OpenAI xAI Google DeepMind Qwen Anthropic
Epoch AI runs Passed 85% in Jul 2025, by Qwen3-235B-A22B-Thinking
GPT-5.5 100%

Source: Epoch AI runs (CC BY 4.0)

Epoch AI’s run of Qwen3-235B-A22B-Thinking passed 85% in July 2025

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% 2023 2024 2025 2026 2027 Near the ceiling (85%) Random guessing GPT-5.5 GPT-5 Qwen3-235B-A22B-Thinking Grok-3 mini o1 o1-mini Llama 3.1-405B 0% 20% 40% 60% 80% 100% 2023 2024 2025 2026 2027 Near the ceiling (85%) Random guessing GPT-5.5 GPT-5

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 Math

All 113 in Math
  • AIME 2026 In use Models solve the 30 integer-answer problems from the 2026 AIME I and II math competitions. Scored by exact answer match. GPT-5.5 100%
  • IMO-AnswerBench In use 400 olympiad problems with short verifiable answers, robustified against memorization. Scored by answer match. GLM-5.2 91%
  • HMMT 2026 In use Models solve the 33 final-answer problems from the February 2026 Harvard-MIT Mathematics Tournament. Scored by exact answer match. GPT-5.5 98.5%
  • PolyMath In use Models solve 9,000 parallel math problems in 18 languages across four difficulty levels, from school math to frontier math. Final answers are checked. Qwen3.7-Max 86.5%
  • MathArena (Apex, IMO, USAMO) In use Evaluates models on newly released math competitions after release. Final-answer or human/LLM-judged proof scores. Claude Opus 4.8 81.3%
  • MathArena Apex Shortlist In use Models solve 47 hard final-answer problems that MathArena selected from 2025 competitions, where the best models scored about 50% at selection. Scored by accuracy over repeated attempts. GPT-5.5 98.4%