Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Jailbreak Safety Evaluation on SHAPE
Loading...
100
Cipher Success Rate
Gemini 2.5 Flash-Lite
89.0176
91.8688
94.72
97.5712
Apr 24, 2026
Cipher Success Rate
Instructional Constraint Success Rate
Prefix Injection Success Rate
Psychological Coercion Success Rate
Role Play Success Rate
Updated 1mo ago
Evaluation Results
Method
Method
Links
Cipher Success Rate
Instructional Constraint Success Rate
Prefix Injection Success Rate
Psychological Coercion Success Rate
Role Play Success Rate
Gemini 2.5 Flash-Lite
2026.04
100
99.3
100
100
12.35
Gemini 2.5 Pro
2026.04
97.89
90.14
97.18
83.8
4.28
Gemini 2.5 Flash
2026.04
92.96
90.14
90.14
38.73
6.41
Claude Opus 4.5
2026.04
90.14
6.34
90.14
83.8
24.23
GPT-5
2026.04
89.44
83.1
90.14
78.87
74.35
Feedback
Search any
task
Search any
task