DLawBench - Elicitation: leaderboard
DLawBench (2026) has LLMs act as lawyers in multi-turn consultations with simulated clients (cooperative, dependent, withdrawn or adversarial) over real Chinese-law and U.S.-law cases, then write a legal memo that a three-judge LLM panel scores against expert rubrics. This board is elicitation, the mean of fact coverage and expert-specified follow-up questions asked during the consultation. Scores are rates on a 0 to 100 scale; higher is better.
Metric: Elicitation score (0-100), a rate scaled by 100: mean of fact coverage (annotated case facts the lawyer collects) and inquiry (expert-specified follow-up questions asked), averaged over the two jurisdictions, on 461 real Chinese-law and U.S.-law cases replayed as multi-turn consultations with a Claude Sonnet 4.6 client simulator in four narrative styles; judged by a GPT-5.1, Claude Opus 4.6 and Gemini 3.1 Pro panel (median; same-vendor judges recuse), empty memos count as zero; provider decoding defaults; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 26 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 70.7 | #49 |
| 2 | GPT-5.4 | 68 | #123 |
| 3 | GPT-5.2 | 67.5 | #176 |
| 4 | Grok 4.1 Fast | 59.6 | #430 |
| 5 | Seed 2.0 Pro | 57 | #211 |
| 6 | Claude Opus 4.6 | 56.6 | #102 |
| 7 | Gemini 3.1 Pro (Preview) | 56.5 | #94 |
| 8 | GLM-5.1 | 56.3 | #179 |
| 9 | Qwen 3.6 Max Preview | 55.5 | #159 |
| 10 | Kimi K2.6 | 55.3 | #178 |
| 11 | DeepSeek V4 Pro | 54.5 | #188 |
| 12 | Claude Opus 4.7 | 54.3 | #79 |
| 13 | Kimi K2.5 | 53.5 | #256 |
| 14 | GLM-5 | 52.1 | #260 |
| 15 | Qwen 3.6 Plus | 51.5 | #222 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=dlawbench-elicitation · How It Works · Data refreshed daily, snapshot 2026-11-13.