From what I understand, the NSP on these SoCs performs better on inferences than any other mobile GPU/neural engine. But yeah probably some very significant li

From what I understand, the NSP on these SoCs performs better on inferences than any other mobile GPU/neural engine. But yeah probably some very significant limitations when it comes to LLMs for this generation anyway. Forget about whatever models Microsoft ships, put whatever ones you want on there.