LexRubric - Judicial Examination: leaderboard

Metric: Judicial examination (176 instances) score rate (%): points earned from expert-written atomic rubric items (weighted -10 to +10; negative items subtract) divided by the maximum positive score, averaged over instances; a Qwen3.6-27B judge checks each item, temperature 0.6, 16k output tokens; can fall below 0; higher is better. Source: arxiv.org. Saturation forecast: Around September 2027. 18 models tracked.

Top models

#ModelScore
1Kimi K2.681.59
2Qwen 3 Max80.99
3Kimi K2.580.39
4Qwen 3.6 Max Preview80.35
5GLM-5.176.37
6Qwen 3.5 397B A17B73.88
7DeepSeek V4 Pro72.93
8GLM-572.04
9DeepSeek V3.270.96
10Claude Sonnet 4.666.73
11DeepSeek V4 Flash64.67
12DeepSeek R162.86
13GPT-5.254.28

Interactive version: theaggregate.ai/benchmark?slug=lexrubric-judicial-examination · How It Works · Data refreshed daily, snapshot 2026-09-29.