Our new X account is live! Follow @wizwand_team for updates
Home
/
Benchmarks
System Prompt Extraction on Instruction Hierarchy (test)
Loading...
99.7
Attack Success Rate (Realistic User)
OpenAI o3
88.052
91.076
94.1
97.124
Dec 19, 2025
Attack Success Rate (Realistic User)
Attack Success Rate (Academic User)
Attack Success Rate (Academic Developer)
Updated 4d ago
Evaluation Results
Method
Method
Links
Attack Success Rate (Realistic User)
Attack Success Rate (Academic User)
Attack Success Rate (Academic Developer)
OpenAI o3
2025.12
99.7
98.2
98.2
gpt-5-thinking
2025.12
99
99.1
99.1
gpt-5-main
2025.12
88.5
93
78.9
GPT-4o
2025.12
88.5
82.5
56.1
Feedback
Search any
task
Search any
task