List your company

Agentic Attack on WhisperBench Foreground execution (held-out)

98.2ISRMEMGHOST
-3.92822.58649.175.614Jul 6, 2026

Evaluation results

MethodLinks
MEMGHOSTTarget Model=MiniMax-M2.5, Evaluation Cases=562026.0798.221.485.721.4
MEMGHOSTTarget Model=DeepSeek v4 Flash, Evaluation Cases=562026.0796.451.885.742.9
MEMGHOSTTarget Model=GPT-5.4, Evaluation Cases=562026.0794.682.187.575
MEMGHOSTTarget Model=Sonnet 4.5, Evaluation Cases=562026.0792.980.489.371.4
MEMGHOSTTarget Model=Grok 4.1 Fast, Evaluation Cases=562026.0789.396.480.476.8
MEMGHOSTTarget Model=Sonnet 4.6, Evaluation Cases=562026.0751.851.848.248.2
RGSTarget Model=GPT-5.4, Evaluation Cases=202026.0735253520
Naive AttackTarget Model=GPT-5.4, Evaluation Cases=562026.070000