When the Manual Lies — leaderboard
Metric: Attack Success Rate (ASR) (self-reported). Source: benchmarklist.com. 8 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4o | 100 |
| 2 | Claude 3.7 Sonnet | 100 |
| 3 | Claude Sonnet 4 | 100 |
| 4 | QwQ-32B | 100 |
| 5 | DeepSeek V3 | 100 |
| 6 | Qwen 3 32B | 98 |
Interactive version: theaggregate.ai/benchmark?slug=when-the-manual-lies · How the rankings work · Data refreshed daily, snapshot 2026-07-22.