Yeah, llm in the cloud is a bad idea it's like the IBM mainframe model from the past. Anyone who makes an llm that runs great locally on normal user hardware and uncensored will do so much better.