Over the past few days, I’ve noticed a significant decline in the quality of GPT’s work — both in text and visual tasks. Previously, even for complex and context-dependent requests, the responses were substantive, with nuances and individual style. Now, everything has been reduced to overly general, “template” phrases — as if the assistant has lost all accumulated context.
When generating images with text, meaningless word sets appear, resembling real words but ultimately useless.
• The issue started around July 12–13, 2025.
• Device, browser, VPN, and cache have all been checked and remain unchanged (everything was working fine before).
• I don’t see widespread complaints on forums, but it feels like the quality has globally worsened.
• There’s a noticeable loss of depth, coherence, and individual tone in the text. Visual text has turned into gibberish.
Is anyone else experiencing something similar?
I would appreciate any feedback or advice, as well as any updates from the development team.
Same here. It just ignores the input image and misses important points when it comes to images, and when it comes to text, it has stopped properly interpreting my requests. Now I have to take it by the hand and guide it like it’s a toddler, and it has become so frustrating that I cancelled my subscription. I’d rather go to a third party that offers all-in-one API access, which is luckily still functional.
I would like to say that regarding the recurring problem with image generation. I noticed that, despite several attempts, the images generated in a manga or manhwa style continue to be distorted and do not comply with what I expect in terms of quality and precision.
In particular, the characters’ faces are often deformed, and the bodies too, everything!
which significantly affects the final result. I tried several times to restart image generation with detailed requests, but the problems persist.
I thank you in advance for your attention to this problem and I hope you can inform me about the evolution of the situation.
