Guess Apple didn't quite grasp how many devs would snap these up for local LLMs. M-series chips are surprisingly good for it.