I believe that frontier-model developers should publish their model cards before commercial release. Transparency is crucial in AI development, and model cards provide valuable information about the capabilities, limitations, and ethical considerations of the models. This practice can foster trust and guide responsible AI deployment.