IMOAnswerBench
Can the model solve difficult problems that require more than factual recall?
Reasoning14 ranked models17 reported values3 reportsHigher is better
Top rankings
One row per model, using its best reported score across effort settings.
| Rank | Model | Best score | Best reported setting |
|---|---|---|---|
| 1 | GLM-5.2Z.ai | 91.0% | max effortZ.ai report1 value · 1 reportJun 16, 2026 · source |
| 2 | Qwen3.7-MaxQwen | 90.0% | Reported configurationZ.ai report1 value · 1 reportJun 16, 2026 · source |
| 3 | DeepSeek V4 ProDeepSeek | 89.8% | Reported configurationZ.ai report1 value · 1 reportJun 16, 2026 · source |
| 4 | Claude Opus 4.5Anthropic | 84.0% | Reported configurationQwen report1 value · 1 reportApr 21, 2026 · source |
| 5 | GLM-5.1Z.ai | 83.8% | Reported configurationZ.ai report1 value · 1 reportJun 16, 2026 · source |
| 6 | Claude Opus 4.8Anthropic | 83.5% | Reported configurationZ.ai report1 value · 1 reportJun 16, 2026 · source |
| 7 | Gemini 3.1 ProGoogle | 81.0% | Reported configurationZ.ai report1 value · 1 reportJun 16, 2026 · source |
| 8 | Qwen3.5-397B-A17BQwen | 80.9% | Reported configurationQwen report1 value · 1 reportApr 21, 2026 · source |
| 9 | Qwen3.6-27BQwen | 80.8% | Reported configurationQwen report1 value · 1 reportApr 21, 2026 · source |
| 10 | Qwen3.5-27BQwen | 79.9% | Reported configurationQwen report2 values · 2 reportsApr 21, 2026 · source |
| 11 | Qwen3.6-35B-A3BQwen | 78.9% | Reported configurationQwen report2 values · 2 reportsApr 21, 2026 · source |
| 12 | Qwen3.5-35B-A3BQwen | 76.8% | Reported configurationQwen report1 value · 1 reportApr 15, 2026 · source |
Effort curve
Every sourced cost-linked effort value for this exact version. Lines connect complete sweeps only.
No cost-linked effort sweep for this version.
Definition and comparison boundarypublic methodology
Can the model solve difficult problems that require more than factual recall?
Multi-step reasoning on academic, mathematical, or abstract problems. Higher is better. The value is the percentage reported in this lab's table.
Harness, tools, prompts, attempt count, and benchmark version can materially change the result. Compare within one reporting context.
Method / source