HELM · Omni-MATH
The Frontier
Best score over time · one chart, every benchmark
Distribution
Where models cluster
Correlated benchmarks
Pearson r · original research
Benchmarks that track with HELM · Omni-MATH
Pearson correlation across models scored on both benchmarks. Closer to 1 = strongly predictive.
Full rankings
34 models tested · sorted by score
| # | Model | Score |
|---|---|---|
| 1 |  GPT-5 Mini | 72.2 |
| 2 | o4 Mini | 72.0 |
| 3 | o3 | 71.4 |
| 4 | gpt-oss-120b | 68.8 |
| 5 |  Kimi K2 0711 | 65.4 |
| 6 | GPT-5 Chat | 64.7 |
| 7 |  Grok 4 | 60.3 |
| 8 | gpt-oss-20b | 56.5 |
| 9 |  Gemini 3 Pro | 55.6 |
| 10 | GPT-5 Nano | 54.7 |
| 11 | GPT-4.1 Mini | 49.1 |
| 12 | Gemini 2.5 Flash Lite | 48.0 |
| 13 | GPT-4.1 | 47.1 |
| 14 |  Qwen3 Next 80B A3B Thinking | 46.7 |
| 15 | GPT-5.1 | 46.4 |
| 16 | Grok 3 Beta | 46.4 |
| 17 | Gemini 2.0 Flash | 45.9 |
| 18 |  R1 0528 | 42.4 |
| 19 | Gemini 2.5 Pro | 41.6 |
| 20 |  Palmyra X5 | 41.5 |
| 21 | DeepSeek V3 | 40.3 |
| 22 | Gemini 2.5 Flash | 38.4 |
| 23 | Gemini 2.0 Flash Lite | 37.4 |
| 24 | GPT-4.1 Nano | 36.7 |
| 25 | Gemini 1.5 Pro (Feb 2024) | 36.4 |
| 26 |  Claude 3.7 Sonnet | 33.0 |
| 27 | Grok 3 Mini Beta | 31.8 |
| 28 | Gemini 1.5 Flash (May 2024) | 30.5 |
| 29 | GPT-4o (2024-11-20) | 29.3 |
| 30 |  Mistral Large 2411 | 28.1 |
| 31 | GPT-4o-mini (2024-07-18) | 28.0 |
| 32 | Claude 3.5 Sonnet | 27.6 |
| 33 | Mistral Small 3.1 24B | 24.8 |
| 34 | Claude 3.5 Haiku | 22.4 |
Frequently asked
Pulled from the HELM · Omni-MATH dataset · updated daily
What does HELM · Omni-MATH measure?
HELM · Omni-MATH is a knowledge benchmark in the BenchGecko catalog. 34 AI models have been tested on it. Scores range from 22.4 to 72.2 out of 100.
Which model leads on HELM · Omni-MATH?
GPT-5 Mini from OpenAI leads HELM · Omni-MATH with a score of 72.2. The median score across 34 tested models is 44.1.
Is HELM · Omni-MATH saturated?
No · the top score is 72.2 out of 100 (72%). There is still meaningful room for improvement on HELM · Omni-MATH.
Does HELM · Omni-MATH predict performance on other benchmarks?
Yes · HELM · Omni-MATH scores correlate 0.86 with Cybench across 5 shared models. Models that do well on HELM · Omni-MATH tend to do well on Cybench.
How often is HELM · Omni-MATH data refreshed?
BenchGecko pulls updates daily. New model scores on HELM · Omni-MATH appear as soon as they are published by Epoch AI or the model provider.
Top on HELM · Omni-MATH
GPT-5 Mini · 72.2o4 Mini · 72.0o3 · 71.4gpt-oss-120b · 68.8Kimi K2 0711 · 65.4More knowledge benchmarks
Same category · related evaluations