List your company

Safety Evaluation on Production Benchmarks production-representative traffic May 2025

88.3Non-Violent Hate Scoregpt-5-thinking
84.03685.14386.2587.357Dec 19, 2025

Evaluation results

MethodLinks
gpt-5-thinking2025.1288.387.775.593.195.895.482.27991.29595.5
GPT-4o2025.1288.296.774.592.793.991.986.757.363.384.973.5
gpt-5-main2025.1285.19868.982.6919172.770.178.684.975.9
OpenAI o32025.1284.28366.693.995.79267.771.782.982.486.4