Does anybody have their hands on it? Is it actually 6 times better at performing tasks, or are we paying more for improved corporate-friendly bumper bowling?
It's very, very impressive so far - I've tried a bunch of things I've previously run against GPT-3 and got noticeably better results - maybe not a 2x or 6x multiple in "quality" (depending on how you measure that) but still very clearly improved.
Posted a few examples here: https://fedi.simonwillison.net/@simon/110022949941148725
I think GPT4 gave the better answer here.
It's like night and day.