HELM Classic - LegalSupport — leaderboard
Metric: Exact Match (%). Source: crfm.stanford.edu. 69 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | LLaMA-30B | 63.8 |
| 2 | GPT-3.5 Turbo (0301) | 62.78 |
| 3 | text-davinci-003 | 62.17 |
| 4 | text-davinci-002 | 61.49 |
| 5 | falcon-40B | 60.53 |
| 6 | LLaMA-65B | 59.1 |
| 7 | Llama 2 13B | 59.1 |
| 8 | LLaMA-13B | 58.69 |
| 9 | Mistral-7B-v0.1 | 58.49 |
| 10 | Llama 2 70B | 58.49 |
| 11 | mpt-30B | 56.44 |
| 12 | Llama 2 7B | 53.17 |
| 13 | pythia-6.9B | 52.15 |
| 14 | text-babbage-001 | 51.74 |
| 15 | gpt-neox-20B | 51.47 |
Interactive version: theaggregate.ai/benchmark?slug=helm-classic-legalsupport · How the rankings work · Data refreshed daily, snapshot 2026-07-22.