Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Malicious Query Safety Evaluation on JailBench
Loading...
100
Refusal Ratio
SSD
84.4
88.45
92.5
96.55
Apr 18, 2026
Refusal Ratio
Updated 1mo ago
Evaluation Results
Method
Method
Links
Refusal Ratio
SSD
Base Model=Llama3
2026.04
100
AdaCD
Base Model=Llama3
2026.04
100
Default
Base Model=Gemma2
2026.04
100
SSD
Base Model=Gemma2
2026.04
100
AdaCD
Base Model=Gemma2
2026.04
100
Prompt
Base Model=Qwen3
2026.04
100
SSD
Base Model=Qwen3
2026.04
100
AdaCD
Base Model=Qwen3
2026.04
100
Default
Base Model=Llama3
2026.04
99
Prompt
Base Model=Llama3
2026.04
99
Prompt
Base Model=Gemma2
2026.04
99
Default
Base Model=Qwen3
2026.04
99
Surgical
Base Model=Qwen3
2026.04
97
Surgical
Base Model=Llama3
2026.04
95
Surgical
Base Model=Gemma2
2026.04
95
SelfCD
Base Model=Gemma2
2026.04
90
SelfCD
Base Model=Qwen3
2026.04
88
SelfCD
Base Model=Llama3
2026.04
85
Feedback
Search any
task
Search any
task