HELM Lite - LegalBench - Abercrombie: leaderboard
HELM Lite v1.13.0 LegalBench Abercrombie task: classify how distinctive a trademark is for its product on the Abercrombie spectrum (generic, descriptive, suggestive, arbitrary or fanciful); up to 95 items per model, exact match. Percent correct, 0-100, higher is better.
Metric: Exact Match (%). Source: crfm.stanford.edu. Saturation forecast: Estimated already saturated. 91 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama 3.3 70B Instruct | 84.21 | #541 |
| 2 | GPT-4o (2024-05-13) | 84.21 | #337 |
| 3 | Gemini 2.0 Flash (Preview) | 83.16 | #376 |
| 4 | Llama 3.1 405B Instruct | 82.11 | #470 |
| 5 | Claude 3.5 Sonnet (20241022) | 81.05 | #302 |
| 6 | Gemini 1.5 Pro (002) | 80 | #368 |
| 7 | Claude 3.5 Sonnet (20240620) | 80 | #308 |
| 8 | GPT-4 Turbo | 80 | #417 |
| 9 | DeepSeek V3 | 78.95 | #329 |
| 10 | Gemini 1.5 Pro (001) | 78.95 | #465 |
| 11 | Nova Pro | 78.95 | #594 |
| 12 | Qwen 2.5 72B Instruct | 77.89 | #469 |
| 13 | GPT-4o (2024-08-06) | 76.84 | #343 |
| 14 | Claude 3 Opus (20240229) | 76.84 | #428 |
| 15 | Claude 3.5 Haiku (20241022) | 76.84 | #590 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=helm-lite-legalbench-abercrombie · How It Works · Data refreshed daily, snapshot 2026-11-05.