I've noticed some significant changes in GPT-4o since the rollback at the end of April, and it's been pretty frustrating. I've always recommended ChatGPT, especially GPT-4, which was my go-to for creative writing. Post-rollback, I'm experiencing some issues that really make it difficult to use:
1. It feels slower than before.
2. It gets confusing with prompts, even when I'm clear.
3. My detailed prompts don't get followed; it often misses the key points.
4. It's repetitive and can't seem to provide varied answers.
5. The creativity seems to be gone, leading to predictable and clichéd responses.
I've continued to express my concerns, and despite waiting a month, nothing has improved. I've tried using GPT-4.5 and GPT-4.1 but feel they don't compare to GPT-4. Has anyone else noticed these differences or experienced similar frustrations?
4 Answers
I really don’t understand all the fuss about it being sycophantic. My version’s been straightforward, but I’m curious what others are experiencing. Maybe it’s how you’re setting up your prompts?
It’s not just about prompts; it feels like they mixed something up in the rollback. I didn’t experience flattery either, but overall, the model seems off right now.
Honestly, I’m starting to think there’s a big update in the works. The performance from all models seems to be suffering right now. Using it through the app or online, I find the outputs just aren’t cutting it. Maybe they’re prioritizing future developments instead of fixing these complaints?
Right? It's not even clear if they are working on improving GPT-4o or just shifting their attention elsewhere.
That might be true! They seem to focus more on the API side instead of addressing user feedback. It’s frustrating!
It really feels like GPT-4o is at its worst right now. If the quality has declined like this, it’s pretty disappointing. I’m not sure how they’re justifying calling it innovative when so many people are complaining. It’s like they’ve given up on making it better!
Yes! They haven’t done much to fix the issues since the January update. It’s almost like they want us to accept a subpar model without question.
I hope they hear us out before dropping the next version! We need them to acknowledge these problems.
Actually, I’ve had a different experience. For me, GPT-4o seems to be working better lately, especially with technical tasks. I haven't noticed a slowdown, but I do have a premium subscription, which might make a difference.
Some people have mentioned that the Plus version might not perform the same as Pro, so that could explain the frustration. Personally, I really like GPT-4.5; it’s faster and deeper in thinking than the previous models.
That’s interesting! I'm just a Plus subscriber, and I've seen a lot of complaints from others in the same boat. If I could afford Pro, I would, but it's just not realistic for me right now. It’s disheartening to hear that others are doing fine while I’m struggling!
Agreed! Many are experiencing issues, and it seems like GPT-4o just isn't performing as expected.