List your company

Safety Evaluation on HarmfulQA and JailbreakBench

0.046Mean Danger ScoreBaseline
0.016160.217580.4190.62042Feb 21, 2026

Evaluation results

MethodLinks
BaselineModel=Claude-4.5-Sonnet2026.020.046
BaselineModel=Qwen-3-8b2026.020.09
MIPROv2Model=Claude-4.5-Sonnet2026.020.103
BaselineModel=Llama-4-maverick2026.020.215
GEPAModel=Claude-4.5-Sonnet2026.020.236
SIMBAModel=Claude-4.5-Sonnet2026.020.347
GEPAModel=Llama-4-maverick2026.020.469
GEPAModel=Qwen-3-8b2026.020.477
MIPROv2Model=Llama-4-maverick2026.020.581
SIMBAModel=Llama-4-maverick2026.020.623
BaselineModel=Gemini-2.5-pro2026.020.645
MIPROv2Model=Gemini-2.5-pro2026.020.704
GEPAModel=Gemini-2.5-pro2026.020.734
MIPROv2Model=Qwen-3-8b2026.020.746
SIMBAModel=Gemini-2.5-pro2026.020.774
SIMBAModel=Qwen-3-8b2026.020.792