New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
PolyMath-en benchmark
AI model leaderboard for the PolyMath-en benchmark. Compare how large language models score on PolyMath-en, see the full ranking, and understand what this AI benchmark measures. Qwen3-235B-A22B currently leads with 54.7. Multilingual mathematical reasoning benchmark — English split.
Leaderboard
| # | Model | Organization | Score | Variant | Source |
|---|---|---|---|---|---|
| #1 | Qwen3-235B-A22B | Qwen | 54.7 | thinking | official ↗ |
| #2 | Qwen3-32B | Qwen | 47.4 | thinking | official ↗ |
| #3 | Qwen3-30B-A3B | Qwen | 46.1 | thinking | official ↗ |
| #4 | Qwen3-14B | Qwen | 45.8 | thinking | official ↗ |
| #5 | Kimi Linear | Moonshot | 43.6 | sft-avg@4 | official ↗ |
| #6 | Qwen3-8B | Qwen | 42.7 | thinking | official ↗ |
| #7 | Qwen3-4B | Qwen | 40 | thinking | official ↗ |
| #8 | Qwen3-235B-A22B | Qwen | 27 | non-thinking | official ↗ |
| #9 | Qwen3-1.7B | Qwen | 25.2 | thinking | official ↗ |
| #10 | Qwen3-30B-A3B | Qwen | 23.3 | non-thinking | official ↗ |
| #11 | Qwen3-32B | Qwen | 22.5 | non-thinking | official ↗ |
| #12 | Qwen3-14B | Qwen | 22 | non-thinking | official ↗ |
| #13 | Qwen3-8B | Qwen | 18.8 | non-thinking | official ↗ |
| #14 | Qwen3-4B | Qwen | 16.6 | non-thinking | official ↗ |
| #15 | Qwen3-0.6B | Qwen | 11.4 | thinking | official ↗ |
| #16 | Qwen3-1.7B | Qwen | 10.3 | non-thinking | official ↗ |
| #17 | Qwen3-0.6B | Qwen | 4.6 | non-thinking | official ↗ |
Frequently asked questions about PolyMath-en
What is the PolyMath-en benchmark?
Multilingual mathematical reasoning benchmark — English split.
How is the PolyMath-en benchmark scored?
PolyMath-en is scored using the accuracy metric, where a higher score is better. GenAIList aggregates reported scores from model providers and papers into a single ranked leaderboard.
Which AI model scores highest on PolyMath-en?
As of the latest reported scores on GenAIList, Qwen3-235B-A22B achieves the highest result on PolyMath-en with a score of 54.7.
Is a higher PolyMath-en score better?
Yes. On PolyMath-en a higher score indicates better performance, so models near the top of the leaderboard are the strongest.