Skip to content

American Invitational Mathematics Examination 2025

The most recent AIME examination, featuring 15 challenging mathematics problems testing olympiad-level mathematical reasoning with integer answers from 000-999.

38 models

Higher is better.

  1. 1GPT-5.2OpenAI100.0
  2. 2Step 3.5 FlashStepFun97.3
  3. 3MAI-Thinking-1Microsoft97.0
  4. 4Kimi K2.5 (Reasoning)Moonshot AI96.1
  5. 5Kimi K2.5Moonshot AI96.1
  6. 6GLM-4.7Z.AI95.7
  7. 7Ternary Bonsai 2 27BPrism ML95.0
  8. 8MiMo-V2-FlashXiaomi94.1
  9. 9o4-mini (high)OpenAI92.7
  10. 10GPT-OSS 120BOpenAI92.5
  11. 11GPT-OSS 20BOpenAI91.7
  12. 12GLM-4.7-FlashZ.AI91.6
  13. 13Nemotron 3 Super 100BNVIDIA90.2
  14. 14Nemotron 3 Nano 30BNVIDIA89.1
  15. 15DeepSeek V3.1 (Reasoning)DeepSeek88.4
  16. 16Sarvam 105BSarvam88.3
  17. 17Claude Sonnet 4.5Anthropic87.0
  18. 18Granite 4.2 8BIBM86.7
  19. 19MiniCPM5-2BOpenBMB86.5
  20. 20Exaone 4.0 32BLG AI Research85.3
  21. 21Nemotron 3 Nano Omni 30B A3BNVIDIA82.1
  22. 22MiniMax M1 80kMiniMax76.9
  23. 23Nemotron Ultra 253BNVIDIA72.5
  24. 24Qwen3 235B 2507Alibaba70.3
  25. 25QwQ-32BAlibaba69.5
  26. 26Magistral MediumMistral64.9
  27. 27Phi-4-reasoningMicrosoft62.9
  28. 28LFM2.5-2.6BLiquid AI51.9
  29. 29DeepSeek V3.1DeepSeek49.8
  30. 30Kimi K2Moonshot AI49.5
  31. 31LFM2.5-8B-A1BLiquid AI42.5
  32. 32MiniCPM5-1BOpenBMB40.4
  33. 33Claude 4 OpusAnthropic33.9
  34. 34Claude 4 SonnetAnthropic33.1
  35. 35Gemini 2.0 FlashGoogle29.7
  36. 36Qwen3 235B-A22BAlibaba24.7
  37. 37Qwen3 32BAlibaba20.2
  38. 38Gemini 1.5 FlashGoogle14.7
Higher is better. Bar colour: blue proprietary, orange open weights, green other.

About this benchmark

Metric
Integer answers 000-999
Tasks
15 problems
Authors
Mathematical Association of America (2025)

Scores are as reported by model developers or evaluators and may use different settings (for example reasoning effort or tools). Compare with care.