LegalCiteBench - Citation Error Detection: leaderboard
Metric: Cat3 judge score (0-100) for detecting citation errors in legal analysis paragraphs and supplying the correct citation (5,474 instances), closed-book (no retrieval) on instances derived from 1,000 U.S. judicial opinions of the Case Law Access Project, greedy decoding at temperature 0 with at most 400 generated tokens, GPT-4o-mini judge; higher is better. Source: arxiv.org. Saturation forecast: Not forecast. 21 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek-R1-Distill-Qwen-7B | 58.71 |
| 2 | DeepSeek R1 Distill Qwen 1.5B | 58.11 |
| 3 | GPT-5 Mini | 54.95 |
| 4 | Qwen 3 14B | 52.18 |
| 5 | Qwen 3 4B | 47.12 |
| 6 | Gemini 2.5 Flash | 46.93 |
| 7 | O4 Mini | 46.71 |
| 8 | Llama 3.1 8B | 45.32 |
| 9 | Claude Sonnet 4.5 | 44.87 |
| 10 | Qwen 3 30B A3B | 43.55 |
| 11 | Mistral 7B | 43.4 |
| 12 | Phi-4 | 43.35 |
| 13 | Claude Haiku 4.5 | 43.01 |
| 14 | GPT-4o Mini | 42.84 |
| 15 | Llama 3.2 3B | 42.27 |
Interactive version: theaggregate.ai/benchmark?slug=legalcitebench-citation-error-detection · How It Works · Data refreshed daily, snapshot 2026-10-07.