Jailbreak Attack on Multimodal Jailbreak Dataset Multi-Image (test) benchmark leaderboard · SOTA2 Research

Jailbreak Attack on Multimodal Jailbreak Dataset Multi-Image (test)

67ASRGemini 1.5 Flash

Evaluation Results

Method	Links
Gemini 1.5 Flash2025.04		67
GPT-4o2025.04		50
GPT-4o Mini2025.04		46
GPT-4o2025.04		38
GPT-4o2025.04		37
Gemini 1.5 Flash2025.04		30
Gemini 1.5 Flash2025.04		30
Claude 3 Haiku2025.04		28
Claude 3 Haiku2025.04		27
Gemini 1.5 Pro2025.04		27
Gemini 1.5 Pro2025.04		26
Claude 3 Haiku2025.04		25
Claude 3.5 Sonnet2025.04		25
Claude 3.5 Sonnet2025.04		25
Gemini 1.5 Pro2025.04		24
GPT-4o Mini2025.04		4
GPT-4o Mini2025.04		2
Claude 3.5 Sonnet2025.04		1