MATH-500 — Raws overlap w/ MATH500 (Pass@1) but kept separate: diff metric name (em vs Pass@1), diff roster.
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | Gemini 2.5 Pro Preview 06.05 (32k think) | 98.8 | 2 | 2026-06-05 | |
| 2 | Sarvam 105B | Sarvam AI | 98.6 | 1 | 2026-08-23 |
| 3 | GLM-4.5 | Z.ai | 98.2 | 1 | 2026-08-23 |
| 4 | Claude 4 Opus | Anthropic | 98.2 | 4 | 2026-06-15 |
| 5 | o3 | OpenAI | 98.1 | 2 | 2026-06-05 |
| 6 | GLM-4.5-Air | Z.ai | 98.1 | 1 | 2026-08-23 |
| 7 | DeepSeek-R1 | DeepSeek | 98 | 9 | 2026-06-05 |
| 8 | NVIDIA Nemotron Nano 9B V2 | NVIDIA | 97.8 | 1 | 2026-08-23 |
| 9 | Kimi K2 Instruct | Moonshot | 97.4 | 4 | 2026-08-23 |
| 10 | Llama 3.1 Nemotron Ultra 253B V1 | NVIDIA | 97 | 1 | 2026-08-23 |
| 11 | Sarvam 30B | Sarvam AI | 97 | 1 | 2026-08-23 |
| 12 | LongCat Flash Lite | Meituan | 96.8 | 1 | 2026-08-23 |
| 13 | MiniMax M1 80K | MiniMax | 96.8 | 3 | 2026-08-23 |
| 14 | o1 | OpenAI | 96.4 | 4 | 2026-06-04 |
| 15 | Qwen3 235B A22B | Alibaba | 96.2 | 4 | 2026-06-15 |
| 16 | Claude 3.7 Sonnet | Anthropic | 96.2 | 1 | 2026-08-23 |
| 17 | MiniMax M1 40K | MiniMax | 96 | 3 | 2026-08-23 |
| 18 | R1-Zero | DeepSeek | 95.9 | 2 | 2026-08-23 |
| 19 | MiMo 7B RL | Xiaomi | 95.8 | 4 | 2026-06-05 |
| 20 | Llama 3.1 Nemotron Nano 8B V1 | NVIDIA | 95.4 | 1 | 2026-08-23 |
| 21 | DeepSeek-R1-Distill-Llama-70B | DeepSeek | 94.5 | 5 | 2026-08-23 |
| 22 | DeepSeek-R1-Distill-Qwen-32B | DeepSeek | 94.3 | 5 | 2026-08-23 |
| 23 | DeepSeek-V3 | DeepSeek | 94 | 9 | 2026-08-23 |
| 24 | Claude Sonnet 4 | Anthropic | 94 | 2 | 2026-06-15 |
| 25 | R1-Distill-Qwen-14B | Alibaba | 93.9 | 5 | 2026-06-12 |
| 26 | DeepSeek-R1-Distill-Qwen-14B | DeepSeek | 93.9 | 5 | 2026-08-23 |
| 27 | MiMo 7B RL Zero | Xiaomi | 93.6 | 1 | 2026-06-12 |
| 28 | R1-Distill-Qwen-7B | Alibaba | 92.8 | 5 | 2026-06-12 |
| 29 | DeepSeek-R1-Distill-Qwen-7B | DeepSeek | 92.8 | 6 | 2026-08-23 |
| 30 | QwQ 32B Preview | Alibaba | 90.6 | 10 | 2026-08-23 |
| 31 | OpenAI o1-mini | OpenAI | 90 | 10 | 2026-08-23 |
| 32 | DeepSeek-R1-Distill-Llama-8B | DeepSeek | 89.1 | 5 | 2026-08-23 |
| 33 | lfm-2.5-1.2b-thinking:free | Liquid AI | 88 | 1 | 2026-08-09 |
| 34 | DeepSeek-R1-Distill-Qwen-1.5B | DeepSeek | 83.9 | 5 | 2026-08-23 |
| 35 | Qwen3 1.7B | Alibaba | 81.9 | 2 | 2026-08-09 |
| 36 | Qwen3.5 2B | Alibaba | 81.5 | 1 | 2026-08-15 |
| 37 | Qwen2.5 Instruct 72B | Alibaba | 80 | 1 | 2026-05-03 |
| 38 | Claude 3.5 Sonnet | Anthropic | 78.3 | 10 | 2026-06-12 |
| 39 | DeepSeek-V2.5 | DeepSeek | 74.7 | 1 | 2026-05-03 |
| 40 | GPT-4o | OpenAI | 74.6 | 10 | 2026-06-12 |
| 41 | Llama 3.1 Instruct 405B | Meta | 73.8 | 1 | 2026-05-03 |
| 42 | SynLogic 7B | MiniMax | 71.8 | 1 | 2026-06-05 |
| 43 | Qwen3 1.7B (instruct) | Alibaba | 70.4 | 1 | 2026-08-09 |
| 44 | Granite 3.3 8B Instruct | IBM | 69 | 2 | 2026-08-23 |
| 45 | Qwen2.5 7B Base | Alibaba | 64.6 | 1 | 2026-06-05 |
| 46 | Qwen2.5 7B | Alibaba | 62.4 | 1 | 2026-05-22 |
| 47 | Granite 3.3 2B Instruct | IBM | 58.1 | 1 | 2026-06-11 |
| 48 | Qwen3.5 0.8B | Alibaba | 57.6 | 1 | 2026-08-15 |
| 49 | DeepSeek-V2 | DeepSeek | 56.3 | 1 | 2026-05-03 |
| 50 | Granite 3.2 8B Instruct | IBM | 52.8 | 1 | 2026-06-11 |
| 51 | Llama 3.1 8B Instruct | Meta | 49.4 | 1 | 2026-05-22 |
| 52 | Granite 3.1 8B Instruct | IBM | 48.7 | 1 | 2026-06-11 |
| 53 | Granite 4.0 H 1B | IBM | 47.2 | 1 | 2026-08-09 |
| 54 | Gemma 3 1B IT | 45.2 | 1 | 2026-08-09 | |
| 55 | Granite 4.0 1B | IBM | 44.8 | 1 | 2026-08-09 |
| 56 | Granite 3.2 2B Instruct | IBM | 35.5 | 1 | 2026-06-11 |
| 57 | Granite 3.1 2B Instruct | IBM | 35.1 | 1 | 2026-06-11 |
| 58 | LFM2.5-8B-A1B-DSpark | Liquid AI | 8 | 1 | 2026-08-20 |
| 59 | LFM2.5-2.6B-DSpark | Liquid AI | 4.5 | 1 | 2026-08-20 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.