As Apple devices feature faster memory and adopt dual Neural Engines, this significantly improves their ability to run denser AI models. Of course, all products have their limitations, but the company states that its iPhones and iPads are able to run 14-billion-active-parameter models, with Macs being able to blow through the roof by running the trillion-parameter models. A Mac Studio cluster is currently capable of running 1.6+ T AI models, while a single Mac Studio is just shy of 500B LLMs In