The Information report on Gemini + Siri answers most of my questions, including deployment:

To maintain Apple’s privacy pledge, the Gemini-based AI will run directly on Apple devices or its private cloud system, which is powered by Apple’s own server chips, rather than running on Google’s servers. Google put significant engineering effort into getting a version of Gemini working on Apple’s servers, according to a person familiar with the partnership talks.

No small thing for Apple to scale this up on their own. We’ll see small improvements in iOS 26.4, with the biggest changes in the fall.