Skip to content

Harvard-MIT Mathematics Tournament February 2025

A February 2025 HMMT slice used in exact-value provider tables for advanced contest-math reasoning.

13 models

Higher is better.

  1. 1GLM-5Z.AI97.5
  2. 2GLM-4.7Z.AI97.1
  3. 3Qwen3.6 PlusAlibaba96.7
  4. 4Kimi K2.5 (Reasoning)Moonshot AI95.4
  5. 5Kimi K2.5Moonshot AI95.4
  6. 6Qwen3.5 397BAlibaba94.8
  7. 7Qwen3.6-27BAlibaba93.8
  8. 8Nemotron 3 Super 100BNVIDIA93.7
  9. 9Claude Opus 4.5Anthropic92.9
  10. 10Qwen3.6-35B-A3BAlibaba90.7
  11. 11Sarvam 105BSarvam85.8
  12. 12Granite 4.2 8BIBM78.3
  13. 13Sarvam 30BSarvam73.3
Higher is better. Bar colour: blue proprietary, orange open weights, green other.

About this benchmark

Metric
Contest mathematics
Tasks
Competition math problems
Authors
Qwen (2025)

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