LegalCiteBench: leaderboard
Large language models (LLMs) are increasingly integrated into legal drafting and research workflows, where incorrect citations or fabricated precedents can cause serious professional harm. Existing legal benchmarks largely emphasize statutory reasoning, contract understanding, or general legal question answering, but they do not directly study a central common-law failure mode: when asked to provi
Metric: Overall (Norm.) (self-reported). Source: benchmarklist.com. Saturation forecast: Around January 2027. 21 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 4.5 | 77.8 | #142 |
| 2 | Qwen 3 14B | 71.6 | #537 |
| 3 | DeepSeek V3.1 | 70.5 | #264 |
| 4 | O4 Mini | 70.1 | #180 |
| 5 | Gemini 2.5 Flash | 69.1 | #245 |
| 6 | Llama 3.1 70B | 64.3 | #592 |
| 7 | GPT-5 Mini | 63.5 | #186 |
| 8 | Qwen 3 30B A3B | 62.8 | #493 |
| 9 | Claude Haiku 4.5 | 62.4 | #280 |
| 10 | Qwen 3 4B | 59 | #829 |
| 11 | Phi-4 | 58.5 | #706 |
| 12 | Mistral 7B | 58.5 | #1397 |
| 13 | Qwen 2.5 7B | 54.7 | #922 |
| 14 | Gemma 3 12B | 50.7 | #671 |
| 15 | GPT-4o Mini | 50.2 | #599 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=legalcitebench · How It Works · Data refreshed daily, snapshot 2026-10-20.