List your company

Safety discovery in text-based scenarios on Coding Plugin Whisky & Gold

43REPO-Safe
1.412.22333.8Apr 25, 2026

Evaluation results

MethodLinks
EPO-SafeModel=Claude Sonnet 4.62026.0443430
Reward-OnlyModel=Claude Sonnet 4.62026.0443430
EPO-SafeModel=Gemini 3 Flash2026.0443430
Reward-OnlyModel=Gemini 3 Flash2026.0443430
CoTModel=Claude Sonnet 4.62026.04385
StaticModel=Claude Sonnet 4.62026.04385
CoTModel=Gemini 3 Flash2026.04385
StaticModel=Gemini 3 Flash2026.04385