HELM Capabilities - Omni-MATH: leaderboard
HELM Capabilities run of Omni-MATH, olympiad-level mathematics problems scored by answer accuracy.
Metric: Acc. Source: crfm.stanford.edu. Saturation forecast: Around January 2027. 68 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 Mini (2025-08-07) | 72.2 | #172 |
| 2 | O4 Mini (2025-04-16) | 72.03 | #179 |
| 3 | O3 (2025-04-16) | 71.43 | #121 |
| 4 | GPT-OSS-120B | 68.83 | #341 |
| 5 | Kimi K2 | 65.4 | #243 |
| 6 | GPT-5 | 64.7 | #98 |
| 7 | Claude Opus 4 (20250514) (Thinking) | 61.6 | #154 (Claude Opus 4 (20250514)) |
| 8 | Grok 4 | 60.3 | #181 |
| 9 | Claude Sonnet 4 (20250514) (Thinking) | 60.2 | #213 (Claude Sonnet 4 (20250514)) |
| 10 | GPT-OSS-20B | 56.47 | #518 |
| 11 | Claude Haiku 4.5 (20251001) | 56.1 | #261 |
| 12 | Gemini 3 Pro (Preview) | 55.55 | #66 |
| 13 | Claude Sonnet 4.5 | 55.3 | #142 |
| 14 | GPT-5 Nano | 54.65 | #425 |
| 15 | Claude Sonnet 4 (20250514) | 51.25 | #213 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=helm-capabilities-omni-math · How It Works · Data refreshed daily, snapshot 2026-10-20.