[seb314]

For running llms _locally_ on Android, there's "pocketpal" (~7tok/s on a pixel 7 pro for some quant of llama 3.2 3B).

(Not sure if it uses ollama though)