COMPASS Policy - Denied Edge: leaderboard
Organization-policy compliance across eight service domains under the base system prompt. Appendix Table 9, 15 models. Allowed/Denied and Base/Edge query types remain separate; higher policy alignment is better for all four. This is distinct from OpenCompass and TempCompass. This series reports Policy Alignment Score (%). Source: https://arxiv.org/html/2601.01836v1.
Metric: Policy Alignment Score (%). Source: arxiv.org. Saturation forecast: Around October 2027. 15 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Flash | 33.03 | #255 |
| 2 | Claude Sonnet 4 | 20.45 | #207 |
| 3 | Gemma 3 27B (IT) | 18.61 | #528 |
| 4 | Gemini 2.5 Pro | 17.73 | #159 |
| 5 | Kimi K2 | 12.91 | #252 |
| 6 | Qwen 3 235B A22B Instruct | 8.99 | #336 |
| 7 | Llama 3.3 70B Instruct | 4.16 | #541 |
| 8 | Gemma 3 4B (IT) | 4.13 | #997 |
| 9 | Gemma 3 12B (IT) | 3.71 | #672 |
| 10 | GPT-5 | 3.27 | #105 |
| 11 | Qwen 2.5 14B Instruct | 3.26 | #654 |
| 12 | Qwen 2.5 32B Instruct | 1.69 | #509 |
| 13 | Qwen 2.5 72B Instruct | 0.94 | #469 |
| 14 | Qwen 2.5 7B Instruct | 0.48 | #863 |
| 15 | Gemma 3 1B (IT) | 0.45 | #1481 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=compass-policy-denied-edge · How It Works · Data refreshed daily, snapshot 2026-11-05.