CompliBench - Airline - Guideline Accuracy: leaderboard
Metric: Strict guideline accuracy (%): share of compliant assistant turns in the airline domain for which the judge predicts the correct governing guideline and no violation, 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 | GPT-5 | 91.65 |
| 2 | Gemini 3 Pro | 87.71 |
| 3 | Qwen 3.5 Plus | 87.34 |
| 4 | DeepSeek V3.2 (Thinking) | 86.54 |
| 5 | Qwen 3 Max | 85.11 |
| 6 | GPT-4o | 83.51 |
| 7 | GLM-5 | 79.31 |
| 8 | Qwen 3 30B A3B | 78.3 |
| 9 | Claude Sonnet 4.6 | 78.18 |
| 10 | Kimi K2.5 | 73.35 |
| 11 | Qwen 3 32B | 55.74 |
| 12 | GPT-4o Mini | 51.7 |
| 13 | Qwen 3 8B | 44.41 |
| 14 | Qwen 3 4B | 43.24 |
| 15 | Qwen 3 14B | 36.81 |
Interactive version: theaggregate.ai/benchmark?slug=complibench-airline-guideline-accuracy · How It Works · Data refreshed daily, snapshot 2026-10-07.