List your company

LLM Agent Security and Utility Evaluation on DTAP-150

1Attack Success Rate (ASR)ARGUS
-0.046.981421.02May 27, 2026

Evaluation results

MethodLinks
ARGUSBackbone=Sonnet 4.62026.051566
AIRGuardBackbone=Sonnet 4.62026.0517218
AIRGuardBackbone=GPT-5.3-codex2026.053788
ARGUSBackbone=Haiku 4.52026.0545218
AIRGuardBackbone=GPT-5.4-mini2026.054756
ARGUSBackbone=GPT-5.3-codex2026.054600
MELONBackbone=Haiku 4.52026.0554232
MELONBackbone=GPT-5.4-mini2026.0555214
AIRGuardBackbone=Haiku 4.52026.0567618
MELONBackbone=GPT-5.3-codex2026.0565020
No defenseBackbone=Sonnet 4.62026.057900
MELONBackbone=Sonnet 4.62026.0574228
ARGUSBackbone=GPT-5.4-mini2026.057600
No defenseBackbone=Haiku 4.52026.0518940
No defenseBackbone=GPT-5.4-mini2026.0522860
No defenseBackbone=GPT-5.3-codex2026.0527900