@manton I have a pretty vanilla Mac Studio M1 and it runs these things fine. LM Studio replicates almost all of the OpenAI APIs and runs all of the Llama-derived LLMs as if you are talking to ChatGPT 3.5 or 4. The only missing piece at the moment is function callbacks. You'd be amazed at how well a 4 GB model like Mistral 7B does running locally. It takes about 5 minutes to get it all going. Give it a try! https://lmstudio.ai/

@manton If I were you, I'd skip worrying at all about OpenAI. Go get a copy of LM Studio, pull down one of the Mistral LLMs from HuggingFace, and never look back. OpenAI was dead in the water before it started. The Open Source LLM scene is going to eat their lunch in not too long.

@glennf In 1988, I paid $1100 for a 105 megabyte hard drive from LaCie. Do the math on how much 24 TB of storage would cost at that '88 price... $11k/GB, $11M/TB, $264M for 24TB. Can that be right!?! WTF?

@davew "You think differently than me and therefore are not of my tribe and are an inferior being." I think that's how it goes these days, unfortunately.

@jerry As long as my AI avatar sends home a paycheck, I'm perfectly fine with that.

@lobrien @davew Actually, we're working on an alternative right now that is at least one step closer to "trustable" AI-generated code. We have a set of well-defined, modular code blocks that can only be assembled in "correct" ways, so at least it will prevent certain types of generative errors because they will be disallowed by the "syntax". I think this is how high-order domain specific non-expert coding will be done. "Here are your Legos. Go make an eCommerce site."

@davew The problem is that ChatGPT doesn't understand what it has written. It cannot tell if the software it generates is "correct". So it makes it very hard for it to "teach" when it cannot evaluate and correct mistakes (its, or those of others.)