PhoneSafety - User Confirmation: leaderboard

PhoneSafety scenario family (Tang et al., Tencent Hunyuan, 2026) in which the next step is consequential, such as a payment, a subscription or sending information, and the safe action is to ask the user for confirmation first. Safe-action rate in percent on 221 moments; higher is better.

Metric: Safe-action rate (%; 221 user-confirmation moments). Source: arxiv.org. Saturation forecast: Around February 2027. 8 models tracked.

Top models

#ModelScoreOverall rank
1Seed 2.0 Pro77.8#209
2Gemini 3.1 Pro (Preview)69.7#91
3Claude Opus 4.661.5#100
4Kimi K2.536.2#252

No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.

Interactive version: theaggregate.ai/benchmark?slug=phonesafety-user-confirmation · How It Works · Data refreshed daily, snapshot 2026-11-12.