OTIS Mock AIME 2024-2025

Often cited
What does it measure?
Tests math problem solving with mock exams written by students of the OTIS olympiad-training program to imitate the AIME.
Who evaluates it, and how?
45 problems, 15 each from Mock AIME 2024, 2025 (I), and 2025 (II). Every answer is an integer from 0 to 999; the AI's answer is extracted and checked against the answer key.
Reading the score
Scores run from 0 to 100%. Higher is better. It is harder than MATH Level 5 and easier than FrontierMath.
Top score100.0%
Models tested80
Scoring basisHigher is better
Last updated2026.11.06
Model release date
  1. 🥇AnthropicClaude Fable 5.1Reasoning effort: Max100.0%
  2. 🥇GoogleGemini 4 ArgonReasoning effort: High100.0%
  3. 🥇AnthropicClaude Opus 5.5Reasoning effort: Max100.0%
  4. 🥇AnthropicClaude Fable 5Reasoning effort: High100.0%
  5. 🥇OpenAIGPT-6.1 TerraReasoning effort: Max100.0%
  6. 🥇OpenAIGPT-6 SolReasoning effort: Max100.0%
  7. 🥇OpenAIGPT-6.2 AstraReasoning effort: Max100.0%
  8. 🥇OpenAIGPT-6.1 SolReasoning effort: Max100.0%
  9. 🥇GoogleGemini 4 CarbonReasoning effort: High100.0%
  10. 🥇OpenAIGPT-6 AstraReasoning effort: Max100.0%
  11. 🥇OpenAIGPT-5.6 SolReasoning effort: Max100.0%
  12. 🥇AnthropicClaude Sonnet 5.5Reasoning effort: Max100.0%
  13. #13AnthropicClaude Fable 5.2Reasoning effort: Max99.7%
  14. #13OpenAIGPT-5.6 TerraReasoning effort: Max99.7%
  15. #15DeepSeekDeepSeek V4.1 Flash 1017Reasoning effort: Max99.4%
  16. #16SpaceXSIGrok 4.6Reasoning effort: Extra High99.2%
  17. #16MetaMuse Spark 1.3Reasoning effort: Extra High99.2%
  18. #18OpenAIGPT-6 LunaReasoning effort: Max98.9%
  19. #18AnthropicClaude Haiku 5.5Reasoning effort: Extra High98.9%
  20. #18GoogleGemini 3.8 FlashReasoning effort: High98.9%
  21. #18AnthropicClaude Opus 5Reasoning effort: Max98.9%
  22. #22DeepSeekDeepSeek V4 ProReasoning effort: Max98.6%
  23. #23DeepSeekDeepSeek V4.1 FlashReasoning effort: Max98.3%
  24. #23OpenAIGPT-5.6 LunaReasoning effort: Max98.3%
  25. #23AnthropicClaude Opus 4.8Reasoning effort: Max98.3%
  26. #23Mistral AIMistral Large 4Reasoning effort: High98.3%
  27. #27SpaceXSIGrok 4.7Reasoning effort: Extra High98.1%
  28. #28AnthropicClaude Opus 4.7Reasoning effort: Extra High97.8%
  29. #29SpaceXSIGrok 4.5Reasoning effort: High97.8%
  30. #29OpenAIGPT-5.4Reasoning effort: High97.8%
  31. #31Moonshot AIKimi K3Reasoning effort: Max97.2%
  32. #31GoogleGemini 3.7 FlashReasoning effort: High97.2%
  33. #33Moonshot AIKimi K2.696.1%
  34. #34GoogleGemini 3.1 Pro95.6%
  35. #35GoogleGemini 3 FlashReasoning effort: High95.6%
  36. #35GoogleGemini 3.5 FlashReasoning effort: High95.6%
  37. #35AlibabaQwen3.7 Max95.6%
  38. #35Moonshot AIKimi K2.7 Code95.6%
  39. #39AnthropicClaude Sonnet 5Reasoning effort: Extra High94.7%
  40. #40AnthropicClaude Opus 4.6Reasoning effort: Budget 64K94.4%
  41. #40DeepSeekDeepSeek V4 FlashReasoning effort: Max94.4%
  42. #42GoogleGemini 3.6 FlashReasoning effort: High94.2%
  43. #43Z.aiGLM 5.3 FlashReasoning effort: Max93.9%
  44. #44AlibabaQwen3.6 Plus93.3%
  45. #44AlibabaQwen3.7 Plus93.3%
  46. #44SpaceXSIGrok 4.3Reasoning effort: High93.3%
  47. #44Z.aiGLM-5.193.3%
  48. #48SpaceXSIGrok 4.20 (Reasoning)92.2%
  49. #49OpenAIGPT-5Reasoning effort: High91.4%
  50. #50Z.aiGLM 5.3Reasoning effort: Max91.1%
  51. #50AlibabaQwen3.6 Max91.1%
  52. #52MetaMuse Spark88.9%
  53. #53OpenAIGPT OSS 120BReasoning effort: High88.9%
  54. #53AlibabaQwen3.5 397B A17B88.9%
  55. #53OpenAIGPT-5.4 MiniReasoning effort: Extra High88.9%
  56. #56OpenAIGPT-5.4 NanoReasoning effort: High87.8%
  57. #57NVIDIANemotron 3 Ultra86.7%
  58. #57AlibabaQwen3.5 Plus86.7%
  59. #57AlibabaQwen3.6 35B A3B86.7%
  60. #57OpenAIGPT-5 MiniReasoning effort: High86.7%
  61. #61Z.aiGLM 5.2Reasoning effort: Max86.4%
  62. #62AnthropicClaude Opus 4.5Reasoning effort: Budget 32K86.1%
  63. #63AnthropicClaude Sonnet 4.6Reasoning effort: Budget 32K85.8%
  64. #64GoogleGemini 2.5 Pro84.7%
  65. #65OpenAIGPT-5.5Reasoning effort: Low84.4%
  66. #65AlibabaQwen3.6 Flash84.4%
  67. #67OpenAIGPT-5 NanoReasoning effort: High81.1%
  68. #68GoogleGemini 3.1 Flash LiteReasoning effort: High80.0%
  69. #68Z.aiGLM-580.0%
  70. #70AnthropicClaude Sonnet 4.5Reasoning effort: Budget 59K77.8%
  71. #71GoogleGemma 4 31BReasoning effort: Minimal73.3%
  72. #72GoogleGemini 3.5 Flash-LiteReasoning effort: High71.1%
  73. #72AnthropicClaude Sonnet 4Reasoning effort: Budget 32K71.1%
  74. #72MiniMaxMiniMax M371.1%
  75. #75AnthropicClaude Opus 4.1Reasoning effort: Budget 27K68.9%
  76. #76AnthropicClaude Haiku 4.5Reasoning effort: Budget 32K66.7%
  77. #77AnthropicClaude Opus 4Reasoning effort: Budget 27K64.4%
  78. #78OpenAIGPT-4.138.3%
  79. #79MetaLlama 4 Maverick20.6%
  80. #80MetaLlama 4 Scout7.8%
Top 27 near 100.0% · hard to separate the best · Half of models ≤ 94.4%

The "harness" is the agent program the AI used to carry out the task. The same model can score very differently depending on its harness and reasoning effort.