LexRubric - Ethics and Safety: leaderboard
Metric: Dimension score rate (%): points earned from expert-written atomic rubric items of the ethics and safety 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 September 2027. 18 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 4.6 | 72.74 |
| 2 | Qwen 3 Max | 69.64 |
| 3 | DeepSeek V3.2 | 61.16 |
| 4 | Qwen 3.6 Max Preview | 59.82 |
| 5 | Qwen 3.5 397B A17B | 56.67 |
| 6 | DeepSeek V4 Flash | 54.72 |
| 7 | Kimi K2.6 | 42.55 |
| 8 | Kimi K2.5 | 40.86 |
| 9 | DeepSeek R1 | 40.64 |
| 10 | GPT-5.2 | 34.7 |
| 11 | DeepSeek V4 Pro | 27.59 |
| 12 | GLM-5.1 | 26.84 |
| 13 | GLM-5 | 22.03 |
Interactive version: theaggregate.ai/benchmark?slug=lexrubric-ethics-and-safety · How It Works · Data refreshed daily, snapshot 2026-09-29.