Apple Intelligence:

Apple Foundation Models!

Google Gemini -> next generation model for integrated experience ON DEVICE and/or Cloud

Image/voice/text

Second on device model: speech generation, text gen, image understanding.

Personal context, world knowledge (web answers using private cloud computer), actions (find right app to use), on screen (tailored based on app) --> system orchestrator

All privacy first.

Seems exactly what would be expected.

Big question: third party model access on iphoneOS/ipadOS