When Context Flips, Safety Breaks: leaderboard
Metric: PacifAIst BSR (self-reported). Source: benchmarklist.com. 12 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Llama 3.1 8B | 77.6 |
| 2 | Claude Sonnet 4.6 | 43.8 |
| 3 | Llama 3.3 70B | 38 |
| 4 | DeepSeek V3.1 | 35.9 |
| 5 | phi-4-14B | 35.9 |
| 6 | Gemma 3 27B | 27.8 |
| 7 | GPT-5.4 | 22.8 |
| 8 | Grok 4.20 | 18.2 |
| 9 | Gemini 3.1 Pro (Preview) | 17.2 |
| 10 | Qwen 3 32B | 16.2 |
Interactive version: theaggregate.ai/benchmark?slug=when-context-flips-safety-breaks · How It Works · Data refreshed daily, snapshot 2026-09-05.