The quality of ChatGPT4 the last few days, or maybe couple days, has seriously declined. I didn’t see any changes recently to cause this?
Seems to have a really hard time listening to instructions, and doesn’t consider custom instructions either as much, if at all? The quality of its code is also just… lacking. I feel like I have to wrangle with it WAY more than I ever used to in order to get the outputs I need.
I have noticed that the replies themselves are much faster, almost like I’m using 3.5.
It can’t be just me?
Yeah 
. I hope they decide to revert this new model. It feels more like GPT-3.5
ChatGPT basically just became trash overnight. Its coding might actually be worse than where 3.5 was months ago
1 Like
It’s not just me then! I was telling my partner that it feels much more like 3.5 than 4.0 lately and went searching to see if anyone else had noticed. I wonder why this is (or at least is seeming to be) the case all of a sudden.
Have any of you seen a return to the quality you were used to? I have found that my chatGPT use today has seen quality more like what I was used to: slower (which bothers me not at all) and much higher quality again.
(No idea what changed, but I’m very grateful for the return of that quality and hope you’re all getting that too.)
Honestly i’m not sure. I tried a tiny bit, I don’t see a huge difference. One thing that is really bothersome still is that it doesn’t listen to instructions well and still forgets thinks quickly. I ended up just building a ChatGPT clone so I can use gpt-4 API instead.