LongJud-Bench - Issue Spotting: leaderboard
Metric: Legal Consultation Score (%; Claude Sonnet 4.6 judge, four 0-10 rubric dimensions per client question, normalized). Source: arxiv.org. Saturation forecast: Around December 2026. 6 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Kimi K2.5 | 67 |
| 2 | GPT-5.2 | 63 |
| 3 | DeepSeek V4 Flash | 62 |
| 4 | Qwen 3.5 Plus | 62 |
| 5 | Qwen 3.5 Flash | 56 |
| 6 | GLM-4.7 | 54 |
Interactive version: theaggregate.ai/benchmark?slug=longjud-bench-issue-spotting · How It Works · Data refreshed daily, snapshot 2026-09-25.