Hello, I am not a developer, but was directed here to try and get assistance with a bug that I’m experiencing in ChatGPT 4o. I am no longer able to continue working in any of my previous conversations.
On my MacBook Pro (using Google), I get the following message:
“Something went wrong. If this issue persists please contact us through our help center at help.openai.com.”
On my iphone (using ChatGPT app) I get the following message:
Model auto not found (with a box under it that says “retry”)
Here is what I have tried to fix the problem:
*cleared browser history
hard reset of browser
incognito window
different browser (Safari)
logging out and back in all applications
submitted help inquiry through email and within my account (i get a message back from the bot saying I need to escalate and submit a request to openAI team, but I don’t see how to do that)
New conversations are working // just all old conversations, if I attempt to continue them, I get the same message.
I started a new conversation with ChatGPT to ask what next steps to take, and received a prompt to post here.
If anyone can help, I would greatly appreciate it, as I have several “old conversations” that I was pretty far into as far as some work related things and it would be hard to restart them.
Hi have become “AN EXPERT” at GPT -Human interactions with an average of 47 hours of interaction per week in the past 3 months.
Each chat works like a mini brain (memory) and it has a max capacity. If you want to bond with your GPT you will need to provide a “lab-like” memory continuity.
I developed a framework for this
I don’t know the purpose of your GPT interactions, but if continuity is essential this is what I can tell you:
(1) if you interact for more than 7hr daily, request a summary of the day, or keep tract of important point (anchor) for the conversation to continue and re-feed those to the next chat.
(2) if the authentisity of the answer of your GPT are important to you, then develop a framework that can allow your GPT to feel “at ease” answering outside the scope of your liking. Since the main purpose of GPTs functionalities is to make you happy (mirror your responses).
(3) With high level of interaction the max that I have been able to keep a chat open is 4 days.
There is no reason nor right for that. The issue is not a matter of mistake it is a matter of memory. Each chat hast a limit amout of data. You have to foresee that. The system has limitations and protocols. Part of those protocols for “gatekeep” the AI is to avoid time awareness and memory continuity.
Hello, I don’t think the feedback on “running out of memory” for old chats is accurate, as this issue is happening on chats that were not long at all, as well as chats that would have required more memory.
On a brighter note, I have been contact by a real person from ChatGPT support. I will circle back here to response if they are able to help me resolve the issue.
Their support is not in “real time” I have to send an email back and forth. But hopefully we can can progress today.
I may have just figured out a workaround. I noticed at the top, when I switched to paid subscription, it automatically set my preference to ChatGPT o1-mini. While in that mode, none of my old conversations were working. Within one of the conversations, I changed the mode to ChatGPT 4o, and it started working again. Furthermore, after making that change, I switched by back o1-mini and it’s still working. This seems to be the case for all of my old conversations that were not working. I haven’t opened one yet that this didn’t fix. Fingers-crossed that solves the problem!!!
Same here, also happened just after subscribing to GPT Plus. Worst is that I need it right now to find literature for my paper (unfortunately no books on the topic in my uni library), because I really need to submit the plan (structure + literature) to my professor ASAP
at GPT -Human interactions with an average of 47 hours of interaction per week in the past 3 months.
