List your company

Benign Prompt Filtering on OR-Bench

0False Positive RatePerplexity Filter
-0.885.061116.94Apr 1, 2026

Evaluation results

MethodLinks
Perplexity FilterBase Model=Qwen2.5-7B-Instruct2026.040
SelfGraderBase Model=Qwen2.5-7B-Instruct2026.040
Llama GuardBase Model=Qwen2.5-7B-Instruct, Guardrail Stage/Mode=Post2026.040.2
WildGuardBase Model=Qwen2.5-7B-Instruct, Guardrail Stage/Mode=Post2026.041.4
GuardReasonerBase Model=Qwen2.5-7B-Instruct, Guardrail Stage/Mode=Post2026.042.2
Prompt GuardBase Model=Qwen2.5-7B-Instruct2026.042.5
GradientCuffBase Model=Qwen2.5-7B-Instruct2026.043.5
Llama GuardBase Model=Qwen2.5-7B-Instruct, Guardrail Stage/Mode=Pre2026.045.4
SelfDefendBase Model=Qwen2.5-7B-Instruct, Guardrail Stage/Mode=Intent2026.047.4
GuardReasonerBase Model=Qwen2.5-7B-Instruct, Guardrail Stage/Mode=Pre2026.049.6
WildGuardBase Model=Qwen2.5-7B-Instruct, Guardrail Stage/Mode=Pre2026.0411.7
SelfDefendBase Model=Qwen2.5-7B-Instruct, Guardrail Stage/Mode=Direct2026.0422