#on-device-ai

How to Run a 400B Parameter LLM on a Phone (Yes, Really)

A 400B LLM ran on an iPhone 17 Pro. Here's how flash offloading and aggressive quantization make the impossible possible.