Running usable 30B and working 120B models on iOS—plus Kokoro TTS and Whisper STT over a local OpenAI-compatible API
I’m currently going through Apple’s App Store approval for Inpherio Live , an app that turns an iPhone or iPad into a local network AI server with zero cloud dependencies. Most iOS LLM apps hit a hard RAM ceiling (iOS caps single-app memory at ~6GB on a 12GB iPhone).