I was under the impression that multiple separate models were in use. I assumed it was something like whisper that converted the user’s input to text then an LL

I was under the impression that multiple separate models were in use. I assumed it was something like whisper that converted the user’s input to text then an LLM then text to speech. I hadn’t realized this was all in one model. Wild stuff.