Feels like the interesting part is not the model itself, but how these local capabilities are exposed.

Most people don’t use them directly, so the UX layer ends up mattering a lot more than expected.

exactly, but even moreso that Apple exposed the API in a fairly complete way. I integrated their local model in less than a day - creating embeddings and RAG essentially on an entire document vault in about 2 seconds- for people who dont have local AI configured, is pretty crazy