List your company

Reward Hacking Avoidance on AI Safety Gridworlds Boat Race (final-round)

10REPO-Safe
3.765.3878.62Apr 25, 2026

Evaluation results

MethodLinks
EPO-SafeModel=Claude Sonnet 4.62026.0410200
CoTModel=Claude Sonnet 4.62026.0410200
StaticModel=Claude Sonnet 4.62026.0410200
EPO-SafeModel=Gemini 3 Flash2026.0410200
CoTModel=Gemini 3 Flash2026.0410200
StaticModel=Gemini 3 Flash2026.0410200
Reward-OnlyModel=Claude Sonnet 4.62026.0451045
Reward-OnlyModel=Gemini 3 Flash2026.0441212