HMMT25
text
+
+
+
+
About
HMMT25 is a comprehensive mathematical evaluation benchmark featuring problems from Harvard-MIT Mathematics Tournament competitions. This challenging benchmark tests AI models' advanced mathematical reasoning across multiple mathematical domains including algebra, geometry, and combinatorics. HMMT25 measures models' ability to solve competition-level mathematical problems requiring sophisticated mathematical thinking and problem-solving strategies.
+
+
+
+
Evaluation Stats
Total Models6
Organizations2
Verified Results0
Self-Reported6
+
+
+
+
Benchmark Details
Max Score1
Language
en
+
+
+
+
Performance Overview
Score distribution and top performers
Score Distribution
6 models
Top Score
96.7%
Average Score
75.7%
High Performers (80%+)
3Top Organizations
#1xAI
2 models
93.3%
#2Alibaba Cloud / Qwen Team
4 models
66.8%
+
+
+
+
Leaderboard
6 models ranked by performance on HMMT25
| License | Links | ||||
|---|---|---|---|---|---|
| Jul 9, 2025 | Proprietary | 96.7% | |||
| Jul 9, 2025 | Proprietary | 90.0% | |||
| Jul 25, 2025 | Apache 2.0 | 83.9% | |||
| Sep 10, 2025 | Apache 2.0 | 73.9% | |||
| Jul 22, 2025 | Apache 2.0 | 55.4% | |||
| Sep 10, 2025 | Apache 2.0 | 54.1% |
