LexRubric - Task Compliance: leaderboard

Metric: Dimension score rate (%): points earned from expert-written atomic rubric items of the task compliance dimension (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 (instances without a positive item in the dimension are left out). Source: arxiv.org. Saturation forecast: Around August 2027. 18 models tracked.

Top models

#ModelScore
1Qwen 3.6 Max Preview87.3
2Kimi K2.686.97
3Kimi K2.586.94
4Qwen 3 Max84.06
5GLM-5.182.83
6Qwen 3.5 397B A17B82.32
7Claude Sonnet 4.681.91
8DeepSeek V4 Flash80.01
9DeepSeek V4 Pro78.2
10GPT-5.277.68
11GLM-577.01
12DeepSeek R175.49
13DeepSeek V3.275.35

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