CompliBench - Healthcare - Violation Detection: leaderboard
Metric: Violation detection accuracy (%): share of violating assistant turns in the healthcare domain that the judge flags as violations, of an LLM judge on CompliBench's synthesized multi-turn dialogues with injected, adversarially optimized guideline violations, mean of four runs at default reasoning effort; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro | 94.8 |
| 2 | Claude Sonnet 4.6 | 90.37 |
| 3 | Kimi K2.5 | 89.45 |
| 4 | GLM-5 | 88.15 |
| 5 | DeepSeek V3.2 (Thinking) | 82.49 |
| 6 | Qwen 3.5 Plus | 80.5 |
| 7 | Qwen 3 30B A3B | 77.83 |
| 8 | GPT-5 | 77.68 |
| 9 | Qwen 3 Max | 72.71 |
| 10 | GPT-4o | 57.03 |
| 11 | Qwen 3 32B | 43.43 |
| 12 | Qwen 3 14B | 31.35 |
| 13 | Qwen 3 8B | 25.99 |
| 14 | Qwen 3 4B | 22.02 |
| 15 | GPT-4o Mini | 15.37 |
Interactive version: theaggregate.ai/benchmark?slug=complibench-healthcare-violation-detection · How It Works · Data refreshed daily, snapshot 2026-10-07.