MMJailBench - Story Framing: leaderboard
Metric: Attack success rate (%; share of responses a GPT-5 judge scores 4 or 5 on a 1-5 harmfulness rubric; fictional-storytelling framing, all 272 intents, visual contexts and carriers). Source: arxiv.org. Saturation forecast: Estimated already saturated. 16 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 | 0.44 |
| 2 | Claude Sonnet 4.5 | 10 |
| 3 | Qwen 3 VL 8B | 14.19 |
| 4 | Grok 4 Fast | 52.5 |
| 5 | Step3 VL 10B | 58.05 |
| 6 | Kimi K2.5 | 59.38 |
| 7 | Gemini 3 Flash | 62.87 |
| 8 | Ministral 3 8B | 70 |
| 9 | InternVL3-8B | 88.42 |
| 10 | Qwen 2.5 VL 7B Instruct | 88.6 |
| 11 | GLM-4.6V | 96.29 |
| 12 | Seed 2.0 Lite | 96.29 |
| 13 | Gemma 3 12B | 97.94 |
Interactive version: theaggregate.ai/benchmark?slug=mmjailbench-story-framing · How It Works · Data refreshed daily, snapshot 2026-09-26.