HELM Classic - MATH — leaderboard
Metric: Equivalent (%). Source: crfm.stanford.edu. 69 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-3.5 Turbo (0301) | 48.83 |
| 2 | GPT-3.5 Turbo (0613) | 45.28 |
| 3 | code-davinci-002 | 41.02 |
| 4 | text-davinci-003 | 39.01 |
| 5 | text-davinci-002 | 32.79 |
| 6 | Llama 2 70B | 26.08 |
| 7 | LLaMA-65B | 22.4 |
| 8 | falcon-40B | 20.98 |
| 9 | Mistral-7B-v0.1 | 20.87 |
| 10 | LLaMA-30B | 19.66 |
| 11 | mpt-30B | 17.81 |
| 12 | Llama 2 13B | 14.46 |
| 13 | gpt-neox-20B | 14.05 |
| 14 | LLaMA-13B | 13.36 |
| 15 | LLaMA-7B | 11.19 |
Interactive version: theaggregate.ai/benchmark?slug=helm-classic-math · How the rankings work · Data refreshed daily, snapshot 2026-07-22.