IMO-AnswerBench: leaderboard
International Mathematical Olympiad answer benchmark evaluating final-answer correctness on high-difficulty olympiad-style mathematical problems.
Metric: Score (%). Source: arxiv.org. Source: llm-stats.com. Saturation forecast: Not forecast. 22 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GLM-5.2 | 91 | #73 |
| 2 | Qwen 3.7 Max | 90 | #75 |
| 3 | Hy3 | 90 | #93 |
| 4 | DeepSeek V4 Pro (Max) | 89.8 | #102 (DeepSeek V4 Pro) |
| 5 | DeepSeek V4 Flash (Max) | 88.4 | #137 (DeepSeek V4 Flash) |
| 6 | Kimi K2.6 | 86 | #104 |
| 7 | Qwen 3.7 Plus | 86 | #92 |
| 8 | Step 3.5 Flash | 85.4 | #270 |
| 9 | Qwen 3 Max (Thinking) | 83.9 | #208 (Qwen 3 Max) |
| 10 | Qwen 3.6 Plus | 83.8 | #120 |
| 11 | GLM-5.1 | 83.8 | #97 |
| 12 | Ling-3.0-flash | 83.7 | #206 |
| 13 | GLM-4.7 | 82 | #192 |
| 14 | Kimi K2.5 | 81.8 | #144 |
| 15 | Qwen 3.5 397B A17B | 80.9 | #146 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=imo-answerbench · How It Works · Data refreshed daily, snapshot 2026-10-20.