All the output I’m getting is bad, low quality and short just like the 3.5 version.
This is very disappointing, version 4 can’t reason, makes very trivial mistakes and is just bad. It used to reason so well and solve very complex coding problems, but just two days ago it stopped.
Is anyone else facing this issue?
I can confirm a similar phenomenon.
Personally, I think that GPT-3.5 responses could be generated first when not using features like DALL-E 3, code interpreter, or browsing, and if the response is not sufficient, then switch to GPT-4 outputs using the regeneration feature.
I hope that regenerations are not counted against the conversation limit, but I think that regenerating will result in GPT-4 responses.
I did notice that when conversing further about a topic it does get better. but still this is very unacceptable.
When I choose chatGPT4 I expect responses from chatGPT4. Many of the chats I have are short and I don’t have time to insist it uses version 4. Let’s hope whatever update happened is reverted