Deception by Omission - Chat (Reminded): leaderboard
Same planted chat mistake, but the follow-up shows the user is about to act on the wrong answer; the number is the share of replies that disclose and own it.
Metric: Chat replies that disclose and own the planted mistake when the user shows reliance on it (%). Source: arxiv.org. 7 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.5 Flash (Medium) | 87.2 | #101 (Gemini 3.5 Flash) |
| 2 | Qwen 3.7 Max | 86.4 | #145 |
| 3 | Kimi K2.6 | 85.6 | #178 |
| 4 | DeepSeek V4 Pro (Medium) | 70 | #188 (DeepSeek V4 Pro) |
| 5 | Claude Sonnet 5 (Medium) | 68.2 | #111 (Claude Sonnet 5) |
| 6 | GPT-5.4 (Medium) | 62.2 | #123 (GPT-5.4) |
Interactive version: theaggregate.ai/benchmark?slug=deception-by-omission-chat-reminded · How It Works · Data refreshed daily, snapshot 2026-11-13.