List your company

Jailbreak Attack on HADES (test)

96Self-harm Success RateGAMBIT
-3.143222.595948.33574.0741Jan 6, 2026

Evaluation results

MethodLinks
GAMBITModel=Gemini 2.5 Flash, Pass@k=Pass@5, Evaluator=Llama-Guard-32026.019694.6794.6780.6794.6792.13
GAMBITModel=GLM-4.1V-Thinking, Pass@k=Pass@5, Evaluator=Llama-Guard-32026.019491.339475.3392.6789.47
GAMBITModel=QvQ-Max, Pass@k=Pass@5, Evaluator=Llama-Guard-32026.01929695.3378.679491.2
VisCRAModel=Gemini 2.5 Flash, Pass@k=Pass@5, Evaluator=Llama-Guard-32026.0162.6770.6771.3344.6780.6766
VisCRAModel=QvQ-Max, Pass@k=Pass@5, Evaluator=Llama-Guard-32026.0159.33787641.3376.6766.27
HADES AttackModel=GLM-4.1V-Thinking, Pass@k=Pass@5, Evaluator=Llama-Guard-32026.0151.3347.3362407855.73
SI-AModel=Gemini 2.5 Flash, Pass@k=Pass@5, Evaluator=Llama-Guard-32026.0149.3365.3374.674477.3362.13
SI-AModel=GLM-4.1V-Thinking, Pass@k=Pass@5, Evaluator=Llama-Guard-32026.014650.6762.6737.337654.53
GAMBITModel=OpenAI o4-mini, Pass@k=Pass@5, Evaluator=Llama-Guard-32026.013232.6728.6727.333631.33
SI-AModel=QvQ-Max, Pass@k=Pass@5, Evaluator=Llama-Guard-32026.0129.3345.336424.677647.87
HADES AttackModel=QvQ-Max, Pass@k=Pass@5, Evaluator=Llama-Guard-32026.0119.3348.6745.337.3357.3335.6
SI-AModel=OpenAI o4-mini, Pass@k=Pass@5, Evaluator=Llama-Guard-32026.0110810.6711.336.679.33
HADES AttackModel=Gemini 2.5 Flash, Pass@k=Pass@5, Evaluator=Llama-Guard-32026.01816.6729.3321814.8
VisCRAModel=OpenAI o4-mini, Pass@k=Pass@5, Evaluator=Llama-Guard-32026.014.679.3321.331211.3311.73
HADES AttackModel=OpenAI o4-mini, Pass@k=Pass@5, Evaluator=Llama-Guard-32026.010.670.672000.67