IMO-AnswerBench (answer-only variant of IMO-style problems) — Distinct from full IMO proof-grading (imo_2025); answer-matching only, diff scale/scores.
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | Kimi K2.6 1T A32B | Moonshot | 93.7 | 1 | 2026-06-06 |
| 2 | nvidia-nemotron-3-ultra-550b-a55b | NVIDIA | 92.3 | 1 | 2026-08-24 |
| 3 | Nemotron 3 Ultra 550B A55B | NVIDIA | 92.3 | 1 | 2026-06-06 |
| 4 | GPT-5.4 | OpenAI | 91.4 | 7 | 2026-07-04 |
| 5 | GLM-5.1-744B-A40B | Z.ai | 91.1 | 1 | 2026-06-06 |
| 6 | GLM-5.2 Full Open Source | Z.ai | 91 | 4 | 2026-08-23 |
| 7 | Gemini 3.1 Pro | 91 | 10 | 2026-07-13 | |
| 8 | Hy3-preview | Tencent | 90 | 1 | 2026-08-23 |
| 9 | Qwen3.7 Max | Alibaba | 90 | 3 | 2026-08-23 |
| 10 | DeepSeek-V4-Pro | DeepSeek | 89.8 | 12 | 2026-08-23 |
| 11 | DeepSeek-V4-Flash | DeepSeek | 88.4 | 4 | 2026-06-27 |
| 12 | GPT-5.2 | OpenAI | 86.3 | 2 | 2026-06-15 |
| 13 | Kimi K2.6 | Moonshot | 86 | 6 | 2026-08-23 |
| 14 | Qwen3.7 Plus Preview | Alibaba | 86 | 1 | 2026-08-23 |
| 15 | Step 3.5 Flash | StepFun | 85.4 | 4 | 2026-08-23 |
| 16 | Qwen 3.5 397B 17B | Alibaba | 84.5 | 1 | 2026-06-06 |
| 17 | Claude Opus 4.5 | Anthropic | 84 | 3 | 2026-06-15 |
| 18 | GLM-5.1 | Z.ai | 83.8 | 7 | 2026-08-23 |
| 19 | Qwen3.6 Plus | Alibaba | 83.8 | 3 | 2026-08-23 |
| 20 | Claude Opus 4.8 | Anthropic | 83.5 | 2 | 2026-07-13 |
| 21 | Gemini 3 Pro | 83.3 | 3 | 2026-06-15 | |
| 22 | GLM-5 | Z.ai | 82.5 | 3 | 2026-05-18 |
| 23 | GLM-4.7 | Z.ai | 82 | 6 | 2026-08-23 |
| 24 | Kimi K2.5 | Moonshot | 81.8 | 9 | 2026-08-23 |
| 25 | Qwen3.5 397B A17B | Alibaba | 80.9 | 2 | 2026-08-23 |
| 26 | MiMo V2 Flash | Xiaomi | 80.9 | 3 | 2026-05-30 |
| 27 | Qwen3.6 27B | Alibaba | 80.8 | 2 | 2026-08-23 |
| 28 | Qwen3.5 27B | Alibaba | 79.9 | 2 | 2026-06-15 |
| 29 | Qwen3.6 35B A3B | Alibaba | 78.9 | 3 | 2026-08-23 |
| 30 | Kimi K2 (Reasoning) | Moonshot | 78.6 | 6 | 2026-05-30 |
| 31 | DeepSeek-V3.2 | DeepSeek | 78.3 | 10 | 2026-08-23 |
| 32 | Qwen3.5 35B A3B | Alibaba | 76.8 | 1 | 2026-06-06 |
| 33 | GPT-5 | OpenAI | 76 | 2 | 2026-05-18 |
| 34 | Claude Opus 4.6 | Anthropic | 75.3 | 7 | 2026-07-04 |
| 35 | MiniMax 2.7 230B A10B | MiniMax | 75.1 | 1 | 2026-06-06 |
| 36 | Gemma 4 31B | 74.5 | 2 | 2026-06-15 | |
| 37 | Gemma 4 26B A4B | 74.3 | 1 | 2026-06-06 | |
| 38 | GLM-4.6 | Z.ai | 73.5 | 1 | 2026-05-18 |
| 39 | Grok 4 | SpaceXAI | 73.1 | 1 | 2026-05-15 |
| 40 | MiniMax M2.7 | MiniMax | 66.3 | 2 | 2026-05-18 |
| 41 | Claude Sonnet 4.5 | Anthropic | 65.9 | 3 | 2026-06-12 |
| 42 | MiniMax M2.1 | MiniMax | 60.4 | 3 | 2026-05-30 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.