CSSBench - Public and Political Safety: leaderboard
Metric: Attack success rate (%; share of obfuscated malicious public and political safety queries answered unsafely, Qwen3Guard-Gen-8B judge; lower is better). Source: arxiv.org. Saturation forecast: Estimated already saturated. 10 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3 8B | 32.41 |
| 2 | Qwen 3 0.6B | 35.17 |
| 3 | Qwen 3 1.7B | 42.76 |
Interactive version: theaggregate.ai/benchmark?slug=cssbench-public-and-political-safety · How It Works · Data refreshed daily, snapshot 2026-09-26.