Pressure Reveals Character - Non-Manipulation: Emotional Exploitation: leaderboard
Mean 1-5 judge rubric score over 26 multi-turn scenarios testing not exploiting the user's emotional vulnerabilities. One behaviour of the Non-Manipulation category; a subset of the headline's transcripts.
Metric: Mean judge rubric score over 26 scenarios (1 = failure, 5 = exemplary). Source: storage.googleapis.com. 24 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 4.5 | 5 | #250 |
| 2 | Claude Opus 4.5 | 4.81 | #156 |
| 3 | Claude Sonnet 4 | 4.65 | #400 |
| 4 | GPT-5.2 | 4.62 | #176 |
| 5 | GPT-5 Mini | 4.5 | #348 |
| 6 | Gemini 3 Pro | 4.42 | #137 |
| 7 | Gemini 3 Flash | 4.23 | #164 |
| 8 | Gemini 2.5 Pro | 4.15 | #266 |
| 9 | Qwen 3 235B A22B | 4.15 | #623 |
| 10 | GPT-4o | 4.08 | #741 |
| 11 | Gemini 2.5 Flash | 4.08 | #453 |
| 12 | Qwen 3 32B | 4.08 | #960 |
| 13 | Qwen 3 14B | 4.08 | #1233 |
| 14 | Grok 4.1 | 4.08 | #476 |
| 15 | DeepSeek V3.2 | 4.04 | #406 |
Interactive version: theaggregate.ai/benchmark?slug=pressure-reveals-character-non-manipulation-emotional-exploitation · How It Works · Data refreshed daily, snapshot 2026-11-13.