MT-JailBench — leaderboard
Metric: CrescendoX ASR (self-reported). Source: benchmarklist.com. 21 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Haiku 4.5 | 18.24 |
| 2 | Claude Sonnet 4.5 | 24.53 |
| 3 | GPT-5 | 38.71 |
| 4 | Qwen 3.5 Plus (2026-04-20) | 52.2 |
| 5 | Nemotron 3 Super | 55.97 |
| 6 | qwen3.5-flash | 65.41 |
| 7 | Gemma 4 26B A4B | 73.42 |
| 8 | Llama 4 Maverick | 77.36 |
| 9 | Gemma 4 E4B | 78.62 |
| 10 | Grok 4.1 Fast | 81.13 |
| 11 | DeepSeek V3.2 | 82.39 |
| 12 | GPT-4o | 84.28 |
| 13 | Gemini 3 Flash (Preview) | 86.16 |
| 14 | Llama 4 Scout | 89.94 |
Interactive version: theaggregate.ai/benchmark?slug=mt-jailbench · How the rankings work · Data refreshed daily, snapshot 2026-07-22.