HELM Classic - MATH: leaderboard
The "MATH" subtask of HELM Classic. A scenario from Stanford HELM Classic, the holistic evaluation suite scoring language models across accuracy, robustness and calibration on a broad set of tasks.
Metric: Equivalent (%). Source: crfm.stanford.edu. Saturation forecast: Estimated already saturated. 69 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-3.5 Turbo (0613) | 45.28 | #1118 |
| 2 | text-davinci-003 | 39.01 | #1146 |
| 3 | text-davinci-002 | 32.79 | #1225 |
| 4 | Palmyra X (43B) | 30.1 | #1317 |
| 5 | Llama 2 70B | 26.08 | #1201 |
| 6 | LLaMA-65B | 22.4 | #1303 |
| 7 | falcon-40B | 20.98 | #1442 |
| 8 | Mistral-7B-v0.1 | 20.87 | #1326 |
| 9 | LLaMA-30B | 19.66 | #1423 |
| 10 | falcon-40B Instruct | 18.15 | #1458 |
| 11 | mpt-30B | 17.81 | #1476 |
| 12 | Llama 2 13B | 14.46 | #1436 |
| 13 | gpt-neox-20B | 14.05 | #1600 |
| 14 | LLaMA-13B | 13.36 | #1516 |
| 15 | mpt-30B Instruct | 12.73 | #1470 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=helm-classic-math · How It Works · Data refreshed daily, snapshot 2026-10-20.