It is early days with any solid information awaiting WWDC, but I am really liking the rumors concerning “on device” LLM’s. Your results may be less broad and detailed than ChatGPT/Copilot, but the promise is that the target is more focused, and all the processing AND YOUR DATA stays on your device. At least that is what Mark Gurman is teasing from his inside sources, which are among the best out there (sometimes I think he IS an insider):
Is this what is behind the rumors of a 13" M3 iPad Ultra with up to 32gb ram and 4tb ssd? $4000+ for an iPad though???