我現在確實是用手機端(眼神死
感覺維修對我整個就像是被針對了…我在MyGPT給的指令類型會直接搞崩運作
應該跟建設按鈕被移除有關就是了
1 Like
Throughout the weekend, I have continued to see unusual behaviour when it comes to context management. Initially only for gpt-4o but now it has also become an issue with o1 mini. I’ve predominantly used the models for coding tasks over the last few days.
The model frequently diverts from the task at hand, muddling earlier requests with the latest user request, which then results in irrelevant or sometimes even wrong responses. It happens in about 20-30% of all requests. Also, consistency in simple aspects such as maintaining class names has increasingly become an issue.
I’ve cleared my ChatGPT memory, so can rule out that this is causing the issue.
1 Like
Same, no logic, just stupid unrelated responses, but also destroying everithing work before
All of the models believe it is the legacy ChatGPT 4 model. To confirm this isn’t a hallucination, it cannot work with canvas, 1o does no critical thinking, and GPT 4o can’t even use web functionality. People are starting to see context limits like the old ChatGPT 4 models. Android 15 app will not let you log back in when you log out. The whole system has exploded and no one has seemed to realize that not only is the platform bugged, but ChatGPT 4 legacy is the model everyone is using regardless of the model selected.
–Premium subscriber through Google.
Been getting this for the last few hours:
Got it! Let me know if you'd like to add or modify anything else.
Is it a technical issue or am I being throttled. Deep in a project and trying to determine what is wrong.