GPT 4 quality drastically reduced?

I have used GPT 4 almost daily for over a year for coding. For the past weeks, it has become virtually unusable; what on earth is going on? I noticed that it is much faster (similar to GPT3.5), but the quality of the response is currently even lower than that of GPT 3.5.

I used to get detailed answers with a lot of attention to context, and it now just throws any random answer, often with errors. For example, I’ll ask it to write a React file and it will import useEffect like useRefEffect. This has never been a hook in React ever, so it’s very confusing to see where this comes from…

I have been close to not using it anymore as, for the past few days it only returns complete nonsense… I’m used to relying on GPT4 for major coding tasks, and I am currently doing it over and over again simply because GPT pays no attention to context or seems to completely change the answer in each message.

Has anyone experienced this as well? is there a way around it?

4 Likes

Same I can stand it for the amount of money I’m paying every month i expect a product which isn’t being downgraded to stupidity. And no way around it

Well, I can’t say that Open AI charges a lot but GPT 4 is getting to the point of being truly useless. For the past days it doesn’t even write code anymore; it just blocks at the first line of code and doesn’t continue.

I messaged OpenAI 3 weeks and even heard back smh.

yes, now they use gpt3.5 instead of 4.0, that’s really disappointing
even I pay 5$ each month for the ChatGPT Plus