I asked more than 25 questions in a thread, and GPT asked if I wanted to continue and downgrade to GPT3.5, so I did
I would like to continue the thread, but cannot go back to GPT4 without starting a fully new chat and losing history
any way to do this?
If you use the playground, you can manually construct/reconstruct the conversation. It would be a lot of copy pasting, but it could maybe get you back to where you were, without using up any of your 25 questions. Just select the “Chat” option under Mode in the playground and use “Add message” to reconstruct it. This is probably the bad option 
but for lack of a better one, right now…
Great idea! I was going to describe a dev route, but you hit the nail on the head.
@01 Keep in mind that GPT3.5 has a lower token limit than GPT4. Token limits are basically how much text can be chewed at any given time all at once, including previous messages of the same conversation. My point is if the content in your messages for a single conversation exceeds what GPT3.5 can handle, it’ll get truncated. I hope this is not the case for you, but if you see shorten responses in comparison to GPT4, you’ll know why.
yeah, I’m trying to figure out if I use the regular chat enough to justify 20 bucks a month.
Right now I have both playground token based account and the gptplus account
it’s on gptplus that gpt4 gets used up and you have to wait. but if you don’t want to wait, it will convert the whole thread to 3.5, but there is no turning back later.
Another downside of playground chats is they aren’t save-able (yet), but you can switch between models up or down (or thats what the UI makes it seem like)
Could you copy and paste (as much) of the conversation as you can, ask for it to be summarized, and then continue from there? It won’t be exactly the same but it’s worth a shot
I could copy/paste, but I have several chats that exceeded the caps, GPT-4 currently has a cap of 25 messages every 3 hours in GPTplus (not playground/api interface). So these are quite long and exceed the character limit.
I read last night about how some people have succeeded in building a browser-side cache to store queries
Great find. Thanks for sharing.
If anything, it leaves me with more questions. Why would OpenAI willingly prevent something that (seemingly) can be so easily implemented? It’s been very frustrating switching to GPT-3.5 and not being able to switch back to GPT-4. I did just try it using a conversation that started with GPT-3.5 and it didn’t do anything. I’m guessing it needs to be a conversation that was started with GPT-4 to work correctly.
Are they worried about “muddying” their results for training? If that was the case, well, RIP. Hopefully they have been recording what model was used for every single message. I’m sure they have. I hope there’s no retribution for doing this. Pushing buttons has already gotten me banned in other OpenAI channels (and I didn’t learn a thing, apparently)
Not being able to switch back to GPT-4 has been brutal.
The last time this happened to me, I waited until it said my message limit had reset (at the hour mark, I think). Is that not normal? I usually just open a new chat and start a different conversation with 3.5 or talk about matters that are adjacent to the original thread of conversation. I don’t actually understand the point of continuing a conversation I started with 4 with 3.5. The reason I started it with 4 at all was becuase 3.5 was insufficient to the task. lol
Anyone got any update on this? I didn’t renew my account for couple of weeks, now all my chats are in GPT-3.5 and no way of going back to GPT-4.
Here I found the solution
Original Article: Reddit - Dive into anything
Video tut: How to switch back from GPT-3.5 to ChatGPT-4 when you reached the limit of 25 messages in 3 hours - YouTube
Hey, not sure if this helps as you all skill is way ahead of mine. What I do is convert the 3.5 thread into a link, then paste the link into a new 4.0 thread and instruct GPT to refresh, retrain, or review the link, or to reference it before every response moving forward.
how do you do it exaclty? I tried to copy the link from old conversation and got the feedback “I apologize for the inconvenience, but I cannot directly access or view external links, including URLs from OpenAI’s platform. However, if you provide a brief summary or description of the content from that link, I’ll do my best to assist you accordingly.”
You could try to copy and paste the conversation into a txt file then print it as pdf and use a chat with pdf tool to resume the conversation. Maybe you can use the txt directly.
Your mileage may vary though as you have to find a plug-in that actually works for your use case, meaning it captures the essence of the conversation.
And when you are at that point you can ask 3.5 to create a summary and use that as a new start for another conversation.
Please allow us in previous chats to select the GPT versions, old chats are stuck in GPT 3.5