So the whole LLM continue to should be saved on-product, but working with it in RAM can be done by means of working with flash memory like a kind of Digital memory, not dissimilar to the way it's completed on macOS for https://appleofeve.co.in/apples-upcoming-release-to-decode/