I wonder if GPT-4’s release was pushed forward to ride the hype that ChatGPT/LLMs have been getting in popular media.

But then again, GPT-3 is almost three years old now, plenty of time to improve upon and train a new model.