HMMT February 2025 — Specific competition-session vintage; never merge across vintage per domain rule.
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | GPT-5.2 | OpenAI | 100 | 2 | 2026-06-23 |
| 2 | Step 3.5 Flash | StepFun | 98.4 | 3 | 2026-05-30 |
| 3 | DeepSeek-V3.2-Speciale | DeepSeek | 97.5 | 1 | 2026-05-11 |
| 4 | Gemini 3 Flash Preview | 97.5 | 1 | 2026-05-11 | |
| 5 | Gemini 3 Pro | 97.5 | 6 | 2026-06-15 | |
| 6 | GLM-4.7 | Z.ai | 97.1 | 4 | 2026-05-30 |
| 7 | GPT OSS 120B | OpenAI | 96.7 | 1 | 2026-06-23 |
| 8 | GPT-5.1 | OpenAI | 96.3 | 1 | 2026-05-18 |
| 9 | Kimi K2.5 | Moonshot | 95.4 | 5 | 2026-06-15 |
| 10 | Qwen3.5 397B A17B | Alibaba | 94.8 | 1 | 2026-06-15 |
| 11 | Qwen3.6 27B | Alibaba | 93.8 | 1 | 2026-06-15 |
| 12 | Kimi K2 (Reasoning) | Moonshot | 93.3 | 7 | 2026-06-13 |
| 13 | Claude Opus 4.5 | Anthropic | 92.9 | 2 | 2026-06-15 |
| 14 | DeepSeek-V3.2 | DeepSeek | 92.5 | 8 | 2026-06-15 |
| 15 | Qwen3.5 27B | Alibaba | 92 | 2 | 2026-06-15 |
| 16 | Gemma 4 26B A4B | 91.7 | 1 | 2026-06-06 | |
| 17 | Qwen3.6 35B A3B | Alibaba | 90.7 | 2 | 2026-06-15 |
| 18 | DeepSeek-V3.2-Exp | DeepSeek | 90 | 1 | 2026-05-30 |
| 19 | GLM-4.6 | Z.ai | 89.2 | 1 | 2026-05-18 |
| 20 | Qwen3.5 35B A3B | Alibaba | 89 | 1 | 2026-06-06 |
| 21 | Gemma 4 31B | 88.7 | 2 | 2026-06-15 | |
| 22 | GPT-5 | OpenAI | 88.3 | 3 | 2026-06-06 |
| 23 | DeepSeek-V3.1 | DeepSeek | 85.8 | 1 | 2026-05-11 |
| 24 | MiMo V2 Flash | Xiaomi | 84.4 | 5 | 2026-06-13 |
| 25 | Gemini 2.5 Pro | 82.5 | 2 | 2026-05-17 | |
| 26 | Claude Sonnet 4.5 | Anthropic | 79.2 | 3 | 2026-06-06 |
| 27 | o3 | OpenAI | 77.5 | 1 | 2026-05-13 |
| 28 | DeepSeek-R1 | DeepSeek | 76.7 | 2 | 2026-05-11 |
| 29 | QED-Nano | LM-Provers | 76.7 | 1 | 2026-06-24 |
| 30 | gpt-oss-20b | OpenAI | 76.7 | 1 | 2026-05-11 |
| 31 | MiniMax M2.1 | MiniMax | 71 | 3 | 2026-05-30 |
| 32 | Gemini 2.5 Flash | 64.2 | 4 | 2026-06-04 | |
| 33 | Qwen3 235B A22B | Alibaba | 62.5 | 3 | 2026-06-25 |
| 34 | DeepSeek-R1 0528-Qwen3-8B | DeepSeek | 61.5 | 2 | 2026-06-04 |
| 35 | Claude 4 Opus | Anthropic | 60 | 1 | 2026-05-11 |
| 36 | Phi 4 Reasoning Plus 14B | Microsoft | 53.6 | 2 | 2026-06-04 |
| 37 | o3-mini | OpenAI | 53.3 | 3 | 2026-06-04 |
| 38 | Qwen3 30B A3B | Alibaba | 50.8 | 1 | 2026-06-25 |
| 39 | Phi 4 Reasoning Plus | Microsoft | 46.7 | 1 | 2026-06-24 |
| 40 | Gemini 2.0 Flash (Non-Reasoning) | 35.8 | 1 | 2026-05-11 | |
| 41 | DeepSeek-R1-Distill-Qwen-32B | DeepSeek | 33.3 | 1 | 2026-05-11 |
| 42 | DeepSeek-R1-Distill-70B | DeepSeek | 33.3 | 1 | 2026-05-11 |
| 43 | DeepSeek-R1-Distill-Qwen-14B | DeepSeek | 31.7 | 1 | 2026-05-11 |
| 44 | Claude 3.7 Sonnet | Anthropic | 31.7 | 1 | 2026-05-11 |
| 45 | DeepSeek-V3 | DeepSeek | 29.2 | 2 | 2026-05-11 |
| 46 | QwQ 32B Preview | Alibaba | 18.3 | 1 | 2026-06-25 |
| 47 | Gemini 2.0 Flash (Reasoning) | 13.3 | 1 | 2026-05-11 | |
| 48 | DeepSeek-R1-Distill-Qwen-1.5B | DeepSeek | 11.7 | 1 | 2026-05-11 |
| 49 | Llama 4 Maverick | Meta | 8.3 | 1 | 2026-05-11 |
| 50 | Gemini 2.0 Pro | 7.5 | 1 | 2026-05-11 | |
| 51 | Claude 3.5 Sonnet | Anthropic | 1.7 | 1 | 2026-05-11 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.