Beyond AIME: leaderboard
Beyond AIME is a difficult mathematical reasoning benchmark testing deeper reasoning chains and harder decomposition than standard AIME-style problem sets.
Metric: Score (percent). Source: www.sarvam.ai. Source: benchmarklist.com. Saturation forecast: Around January 2027. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Sarvam 105B | 69.1 | #409 |
| 2 | Qwen 3 Next 80B A3B (Thinking) | 68 | #320 (Qwen 3 Next 80B A3B) |
| 3 | GLM 4.5 Air | 61.5 | #392 |
| 4 | GPT-OSS-120B | 51 | #341 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=beyond-aime · How It Works · Data refreshed daily, snapshot 2026-10-20.