I've been using GPT4 extensively lately and I'm 60% convinced as well.

I think OpenAI must have a newer model that has no safety guardrails, is allowed to use as much hardware as it wants, and must be highly competitive with the top experts in each field.

After spending some time with the GPT4 API, I'm quite convinced that the thing holding LLMs from being ubiquitous right now is just inferencing cost and safety guardrails.