PolyMATH
multimodal
+
+
+
+
About
PolyMath is a comprehensive multilingual mathematical reasoning benchmark covering 18 languages and 4 difficulty levels, featuring 9,000 high-quality problem samples. This diverse evaluation tests AI models' mathematical capabilities across different linguistic contexts and complexity levels, assessing cross-lingual mathematical understanding, problem-solving skills, and reasoning consistency across various cultural and linguistic mathematical traditions.
+
+
+
+
Evaluation Stats
Total Models4
Organizations1
Verified Results0
Self-Reported4
+
+
+
+
Benchmark Details
Max Score1
Language
en
+
+
+
+
Performance Overview
Score distribution and top performers
Score Distribution
4 models
Top Score
60.1%
Average Score
53.1%
High Performers (80%+)
0Top Organizations
#1Alibaba Cloud / Qwen Team
4 models
53.1%
+
+
+
+
Leaderboard
4 models ranked by performance on PolyMATH
| License | Links | ||||
|---|---|---|---|---|---|
| Jul 25, 2025 | Apache 2.0 | 60.1% | |||
| Sep 10, 2025 | Apache 2.0 | 56.3% | |||
| Jul 22, 2025 | Apache 2.0 | 50.2% | |||
| Sep 10, 2025 | Apache 2.0 | 45.9% |
