HELM Lite - LegalBench - Function of Decision Section: leaderboard
Metric: Exact Match (%). Source: crfm.stanford.edu. 91 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Llama 3 70B | 46.59 |
| 2 | Qwen 2.5 72B Instruct | 46.05 |
| 3 | Gemini 1.5 Pro (001) | 46.05 |
| 4 | Llama 3.1 70B Instruct | 45.78 |
| 5 | Claude 3.5 Sonnet (20240620) | 45.5 |
| 6 | GPT-4 (0613) | 45.23 |
| 7 | Llama 3.2 90B Vision Instruct | 45.23 |
| 8 | Llama 3.1 405B Instruct | 44.96 |
| 9 | Nova Pro | 44.41 |
| 10 | Llama 2 70B | 44.41 |
| 11 | GPT-4o (2024-05-13) | 44.14 |
| 12 | Mixtral 8x22B | 44.14 |
| 13 | Gemini 1.5 Pro (002) | 43.87 |
| 14 | Gemma 2 27B (IT) | 43.87 |
| 15 | Mistral-7B-v0.1 | 43.32 |
Interactive version: theaggregate.ai/benchmark?slug=helm-lite-legalbench-function-of-decision-section · How It Works · Data refreshed daily, snapshot 2026-09-19.