MATH (Hendrycks et al.) — Distinct sampling-metric axis (Maj@1) from Pass@1/EM variants.
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | Qwen3 VL 8B Instruct | Alibaba | 94.6 | 4 | 2026-06-05 |
| 2 | Ministral 3 14B | Mistral | 90.4 | 4 | 2026-06-12 |
| 3 | Qwen3 VL 4B Instruct | Alibaba | 90 | 4 | 2026-06-05 |
| 4 | Ministral 3 8B | Mistral | 87.6 | 4 | 2026-06-12 |
| 5 | Qwen3 14B | Alibaba | 87 | 4 | 2026-06-05 |
| 6 | Gemma 3 12B | 85.4 | 4 | 2026-06-05 | |
| 7 | Ministral 3 3B | Mistral | 83 | 3 | 2026-06-05 |
| 8 | Gemma 3 4B | 75.9 | 4 | 2026-06-05 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.