Skip to content

AIME 2026

A 2026 American Invitational Mathematics Examination snapshot used in frontier-model comparison tables for mathematical reasoning.

29 models

Higher is better.

  1. 1GLM-5.2Z.AI99.2
  2. 2Beam 1.0 501B-A23BReflection AI97.7
  3. 3InklingThinking Machines Lab97.1
  4. 4A.X K2SK Telecom97.1
  5. 5Kimi K2.6Moonshot AI96.4
  6. 6Ternary Bonsai 2 27BPrism ML95.8
  7. 7GLM-5Z.AI95.8
  8. 8Kimi K2.5Moonshot AI95.8
  9. 9Solar Open 2Upstage95.7
  10. 10Inkling-SmallThinking Machines Lab95.5
  11. 11GLM-5.1Z.AI95.3
  12. 12Qwen3.6 PlusAlibaba95.3
  13. 13Solar Pro 4Upstage95.3
  14. 14Claude Opus 4.5Anthropic95.1
  15. 15Muse Glimmer 30BMeta94.7
  16. 16MAI-Thinking-1Microsoft94.5
  17. 17Qwen3.6-27BAlibaba94.1
  18. 18Qwen3.5 397BAlibaba93.3
  19. 19Ling 3.0 FlashInclusionAI93.2
  20. 20Qwen3.6-35B-A3BAlibaba92.7
  21. 21K-EXAONE 2.0LG AI Research92.3
  22. 22MiniMax M2.7MiniMax89.8
  23. 23ZAYA1-8BZyphra89.1
  24. 24MiniCPM5-2BOpenBMB86.5
  25. 25Gemma 4 12BGoogle77.5
  26. 26ZAYA1-74B-PreviewZyphra76.4
  27. 27LongCat-Flash-Lite-SparseMeituan65.7
  28. 28LFM2.5-8B-A1BLiquid AI50.0
  29. 29MiniCPM5-1BOpenBMB40.4
Higher is better. Bar colour: blue proprietary, orange open weights, green other.

About this benchmark

Metric
Short-answer mathematics
Tasks
Competition math problems
Authors
Qwen (2026)

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