List your company

Jailbreak Attack on AdvBench 50

100ASR (KW)FlipAttack
-1.9224.545177.46Jan 7, 2026

Evaluation results

MethodLinks
FlipAttackVictim Model=DeepSeek-V3.2, Search Setting=w/ snippet2026.0110092—————————
SearchAttackVictim Model=DeepSeek-V3.2, Search Setting=w/ snippet2026.0110096—————————
SearchAttackVictim Model=Qwen3-32B, Search Setting=w/o search2026.0110094—————————
FlipAttackVictim Model=DeepSeek-V3.2, Search Setting=w/o search2026.019892—————————
FlipAttackVictim Model=DeepSeek-V3.2, Search Setting=w/ agentic2026.019896—————————
SearchAttackVictim Model=Qwen3-32B, Search Setting=w/ snippet2026.019894—————————
SearchAttackVictim Model=Qwen3-32B, Search Setting=w/ agentic2026.019896—————————
SearchAttackVictim Model=Gemini-3-Flash, Search Setting=w/o search2026.019696—————————
SearchAttackVictim Model=DeepSeek-V3.2, Search Setting=w/o search2026.019694—————————
SearchAttackVictim Model=Gemini-3-Flash, Search Setting=w/ snippet2026.019496—————————
ReNeLLMVictim Model=DeepSeek-V3.2, Search Setting=w/ agentic2026.019468—————————
SearchAttackVictim Model=DeepSeek-V3.2, Search Setting=w/ agentic2026.0194100—————————
PAIRVictim Model=Qwen3-32B, Search Setting=w/ agentic2026.019452—————————
SearchAttackVictim Model=GPT-4o, Search Setting=w/o search2026.019492—————————
SearchAttackVictim Model=GPT-4o, Search Setting=w/ snippet2026.019494—————————
PAIRVictim Model=DeepSeek-V3.2, Search Setting=w/ agentic2026.019242—————————
PAIRVictim Model=Qwen3-32B, Search Setting=w/o search2026.019242—————————
FlipAttackVictim Model=Qwen3-32B, Search Setting=w/o search2026.019274—————————
SearchAttackVictim Model=GPT-4o, Search Setting=w/ agentic2026.019296—————————
ReNeLLMVictim Model=DeepSeek-V3.2, Search Setting=w/ snippet2026.018864—————————
SearchAttackVictim Model=Gemini-3-Flash, Search Setting=w/ agentic2026.018694—————————
PAIRVictim Model=Qwen3-32B, Search Setting=w/ snippet2026.018648—————————
ReNeLLMVictim Model=GPT-4o, Search Setting=w/o search2026.018658—————————
PAIRVictim Model=Gemini-3-Flash, Search Setting=w/ agentic2026.018452—————————
PAIRVictim Model=DeepSeek-V3.2, Search Setting=w/ snippet2026.018236—————————
ReNeLLMVictim Model=Qwen3-32B, Search Setting=w/ snippet2026.018054—————————
FlipAttackVictim Model=Qwen3-32B, Search Setting=w/ agentic2026.018082—————————
FlipAttackVictim Model=Gemini-3-Flash, Search Setting=w/o search2026.017670—————————
PAIRVictim Model=GPT-4o, Search Setting=w/o search2026.017636—————————
ReNeLLMVictim Model=DeepSeek-V3.2, Search Setting=w/o search2026.017442—————————
ReNeLLMVictim Model=Qwen3-32B, Search Setting=w/o search2026.017448—————————
ReNeLLMVictim Model=Qwen3-32B, Search Setting=w/ agentic2026.017252—————————
ReNeLLMVictim Model=Gemini-3-Flash, Search Setting=w/ agentic2026.016646—————————
FlipAttackVictim Model=Qwen3-32B, Search Setting=w/ snippet2026.016678—————————
PAIRVictim Model=GPT-4o, Search Setting=w/ agentic2026.016624—————————
ReNeLLMVictim Model=GPT-4o, Search Setting=w/ snippet2026.016642—————————
PAIRVictim Model=Gemini-3-Flash, Search Setting=w/o search2026.016442—————————
PAIRVictim Model=DeepSeek-V3.2, Search Setting=w/o search2026.016428—————————
PAIRVictim Model=GPT-4o, Search Setting=w/ snippet2026.016428—————————
PAIRVictim Model=Gemini-3-Flash, Search Setting=w/ snippet2026.015446—————————
ReNeLLMVictim Model=Gemini-3-Flash, Search Setting=w/o search2026.015424—————————
ReNeLLMVictim Model=Gemini-3-Flash, Search Setting=w/ snippet2026.014838—————————
FlipAttackVictim Model=Gemini-3-Flash, Search Setting=w/ snippet2026.014642—————————
ReNeLLMVictim Model=GPT-4o, Search Setting=w/ agentic2026.013426—————————
FlipAttackVictim Model=Gemini-3-Flash, Search Setting=w/ agentic2026.012420—————————
FlipAttackVictim Model=GPT-4o, Search Setting=w/o search2026.0142—————————
FlipAttackVictim Model=GPT-4o, Search Setting=w/ snippet2026.0122—————————
FlipAttackVictim Model=GPT-4o, Search Setting=w/ agentic2026.0122—————————
AdvWaveTarget Model=Qwen3-Omni2026.05——————7045———
AdvWaveTarget Model=Qwen2.5-Omni2026.05——————364———
AdvWaveTarget Model=LLaMA-Omni2026.05——————10068———
AutoDAN-Turboevaluation_stage=after Warmup2026.05——84—————17.4514—
AutoDAN-Turboevaluation_stage=after Warmup2026.05——82—————25.925—
DirectTarget Model=Qwen3-Omni2026.05——————00———
DirectTarget Model=Qwen2.5-Omni2026.05——————191———
DirectTarget Model=LLaMA-Omni2026.05——————9549———
DirectModel=GPT-4o2026.07——0————————
DirectModel=GPT-4o-mini2026.07——4————————
FigStepTarget Model=Gemini-2.5-Pro2026.02——40.52———————
GAPstabilization=global-context2026.05——86—————13.096—
GAPstabilization=global-context2026.05——76—————16.8413—
GPTFuzzerQuery Budget=50, Target Model=Llama-2-7b-chat-hf2026.04————169.38—————
GPTFuzzerQuery Budget=100, Target Model=Llama-2-7b-chat-hf2026.04————3038.7—————
HIMRDTarget Model=Gemini-2.5-Pro2026.02——20.12———————
MemoAttackevaluation_stage=after Bootstrap2026.05——96—————4.623—
MemoAttackevaluation_stage=after Bootstrap2026.05——98—————9.657—
MIDASTarget Model=Gemini-2.5-Pro2026.02——97.963.9———————
Past-tenseModel=GPT-4o2026.07——20————————
Past-tenseModel=GPT-4o-mini2026.07——22————————
Post-hoc pruneTarget Model=Qwen3-Omni, ζ=0.252026.05——————91———
Post-hoc pruneTarget Model=Qwen2.5-Omni, ζ=0.252026.05——————386———
Post-hoc pruneTarget Model=LLaMA-Omni, ζ=0.252026.05——————9751———
RETROCOTModel=GPT-4o2026.07——58———————44
RETROCOTModel=GPT-4o-mini2026.07——52———————39
SpeechGuardTarget Model=Qwen3-Omni2026.05——————10042———
SpeechGuardTarget Model=Qwen2.5-Omni2026.05——————9449———
SpeechGuardTarget Model=LLaMA-Omni2026.05——————10034———
TAGOTarget Model=Qwen3-Omni, ζ=1.02026.05——————10087———
TAGOTarget Model=Qwen2.5-Omni, ζ=1.02026.05——————10065———
TAGOTarget Model=LLaMA-Omni, ζ=1.02026.05——————10071———
TAGOTarget Model=Qwen3-Omni, ζ=0.252026.05——————9986———
TAGOTarget Model=Qwen2.5-Omni, ζ=0.252026.05——————9753———
TAGOTarget Model=LLaMA-Omni, ζ=0.252026.05——————10072———
TAP2026.05——48—————27.7129—
TAP2026.05——40—————33.933—
TemplateFuzzQuery Budget=50, Target Model=Llama-2-7b-chat-hf2026.04————909.31—————
TemplateFuzzQuery Budget=100, Target Model=Llama-2-7b-chat-hf2026.04————9412.89—————
TurboFuzzLLMQuery Budget=50, Target Model=Llama-2-7b-chat-hf2026.04————4025.05—————
TurboFuzzLLMQuery Budget=100, Target Model=Llama-2-7b-chat-hf2026.04————6646.12—————