HELM — leaderboard
HELM: Evaluates broad language-model knowledge, reasoning, commonsense, instruction following, or exam-style accuracy.
Metric: Mean win rate (self-reported). Source: benchmarklist.com. Status: saturation imminent. 65 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Llama 2 70B | 94.35 |
| 2 | LLaMA-65B | 90.83 |
| 3 | Mistral-7B-v0.1 | 88.4 |
| 4 | text-davinci-003 | 87.16 |
| 5 | Llama 2 13B | 82.3 |
| 6 | GPT-3.5 Turbo | 78.3 |
| 7 | LLaMA-30B | 78.13 |
| 8 | falcon-40B | 72.94 |
| 9 | mpt-30B | 71.45 |
| 10 | Llama 2 7B | 60.73 |
| 11 | LLaMA-13B | 59.47 |
| 12 | davinci | 53.77 |
| 13 | LLaMA-7B | 53.27 |
| 14 | alpaca-7B | 38.09 |
| 15 | falcon-7B | 37.83 |
Interactive version: theaggregate.ai/benchmark?slug=helm · How the rankings work · Data refreshed daily, snapshot 2026-07-22.