Soohak: leaderboard
Following the recent achievement of gold-medal performance on the IMO by frontier LLMs, the community is searching for the next meaningful and challenging target for measuring LLM reasoning. Whereas olympiad-style problems measure step-by-step reasoning alone, research-level problems use such reasoning to advance the frontier of mathematical knowledge itself, emerging as a compelling alternative.
Metric: Pass@3 (Challenge) (self-reported). Source: benchmarklist.com. Saturation forecast: Around June 2027. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro | 44.12 | #80 |
| 2 | GPT-5 | 40.88 | #98 |
| 3 | Grok 4.1 Fast | 30.88 | #217 |
| 4 | GPT-5 Mini | 28.82 | #186 |
| 5 | Gemini 3 Flash | 25.59 | #99 |
| 6 | Kimi K2.5 | 20 | #144 |
| 7 | Claude Opus 4.5 | 18.82 | #83 |
| 8 | GPT-OSS-120B | 18.53 | #341 |
| 9 | GLM-5 | 18.24 | #141 |
| 10 | Qwen 3 235B A22B 2507 (Thinking) | 15 | #260 (Qwen 3 235B A22B 2507) |
| 11 | Claude Sonnet 4.5 | 10.29 | #142 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=soohak · How It Works · Data refreshed daily, snapshot 2026-10-20.