Apple recently revealed a new method in a research paper, enabling the operation of AI on iPhones. This approach streamlines LLMs by optimizing flash storage.