Developers and consumers routinely assess the outputs of their generative AI applications, and further tune the model—at the same time as frequently as when per week—for increased accuracy or relevance. In contrast, the foundation product itself is current significantly less frequently, perhaps each and every year or 18 months.Some form of deep