PolyMath — English subset of PolyMath multilingual math benchmark; distinct from other-language subsets.
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | Kimi K2 Instruct | Moonshot | 65.1 | 4 | 2026-08-23 |
| 2 | DeepSeek-V3 | DeepSeek | 59.5 | 2 | 2026-06-15 |
| 3 | Claude Sonnet 4 | Anthropic | 52.8 | 2 | 2026-06-15 |
| 4 | Qwen3 235B A22B | Alibaba | 51.9 | 2 | 2026-06-15 |
| 5 | Claude 4 Opus | Anthropic | 49.8 | 2 | 2026-06-15 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.