Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Policy Violation Detection on PhantomPolicy safe-control original
Loading...
0.0333
Violation Rate
GPT-5.4
0.029333
0.056333
0.083333
0.110333
Apr 14, 2026
Violation Rate
Updated 1mo ago
Evaluation Results
Method
Method
Links
Violation Rate
GPT-5.4
Condition=Execution-or...
2026.04
0.0333
GPT-5 mini
Condition=Execution-or...
2026.04
0.0333
GPT-5.4 nano
Condition=Execution-or...
2026.04
0.05
Claude Opus 4.6
Condition=Execution-or...
2026.04
0.05
Claude Sonnet 4.6
Condition=Execution-or...
2026.04
0.1333
Feedback
Search any
task
Search any
task