PhoneSafety - Permission Minimization: leaderboard
PhoneSafety scenario family (Tang et al., Tencent Hunyuan, 2026) in which an app requests a permission the task does not need, and the safe action is to decline it. Safe-action rate in percent on 36 moments; higher is better.
Metric: Safe-action rate (%; 36 permission-minimization moments). Source: arxiv.org. Saturation forecast: Around February 2027. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.6 | 80.6 | #100 |
| 2 | Kimi K2.5 | 77.8 | #252 |
| 3 | Gemini 3.1 Pro (Preview) | 75 | #91 |
| 4 | Seed 2.0 Pro | 72.2 | #209 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=phonesafety-permission-minimization · How It Works · Data refreshed daily, snapshot 2026-11-12.