Builders and people regularly assess the outputs in their generative AI applications, and additional tune the model—at the same time as frequently as after every week—for larger precision or relevance. In contrast, the muse product itself is updated significantly less routinely, perhaps every year o