Queries about selecting GPT-3.5 Option by Default after the Latest GPT-4o update

I’m reaching out to highlight an issue after the latest GPT-4o update that has caused frustration for users like myself. Currently, when encountering the limit with GPT-4o during a chat session, there’s no straightforward option to seamlessly transition back to using GPT 3.5 without starting a new chat. This disrupts the flow of conversations and can be quite inconvenient. The only workaround, as illustrated in the attached screenshot below, involves several manual steps i.e. clicking on the ‘attachment’ button, encountering an error, then changing the model to 3.5, and finally regenerating the answer. This process is not only time-consuming but also frustrating.

Is there a way that allow users to choose their default model preference for chats from the beginning, after the latest GPT-4o update? I mean one that enables users to set GPT 3.5 as their default choice, ensure uninterrupted conversations even after reaching limits with GPT-4o and eliminate the need for manual intervention to switch models mid-conversation.


Having the same problem
I dont want to use GPT-4o (save it for other purpose) instead use GPT-3.5.


Same issue here… It´d be great to be able to set default model to 3.5

I have same issue. Moreover, the switch to 4o happens mid-conversation without any notice.

ChatGPT mentions on starting the conversation with a prompt that specifies the default model. While it acknowledge the model is switched, the generated result is always 4o, until it is manually switched from the reply to 3.5, which later auto-switched to 4o again without notice nor interaction.

And I’m using the web based interface from OpenAI. There’s no selection when logging in, as ChatGPT alluded, and there’s no option to make it default and make it stick.

Same issue here.

One would expect to have the ability to chose the default model to use. Unless OpenAI is researching how users interact with the new model which I believe can still be done if users did have the option to choose regardless.

I’m experiencing same issue where I can’t send more messages. I understand that GPT-3.5 has limitations, but I want to continue using it.

but i find a solution not really it gonna solve but , you can go to settings/Personalization/Memory and turn off the toggle

Where’s this “settings/Personalization/Memory” option found? On the settings page, I just see “General”, “Data Control”, “Security”, and the brand new “Builder Profile”

OK. This is getting ridiculous. Today I’m starting a new conversation. I started the conversation with this prompt:

Please use the GPT-3.5 model for this entire conversation without switching to another model.

While it says it will comply, it will always reply with 4o 1st. Even after manually switched the reply to 3.5, subsequent replies will always be 4o, that have to be manually switched back to 3.5.

Now, I ask how to make the change stick in the conversation. The 4o version have this to say:

You as the user have no control on what the underlying model used

I regret I didn’t screenshot that before I rage-delete the conversation. Is this the direction OpenAI is going?

As for making a conversation stick, I found that only after the 4o responses allotted for the day are used up will the conversation continue in 3.5, as it is requested at the very beginning of the conversation.

Additional info, 5hrs later:

I think Chat-GPT answers about we, as users, having no control on the underlying model used is proven.

I’ve been having a single conversation over the past 5 hrs. Having used up the 4o allotment earlier is making the conversation “defaults” to 3.5.

However, unknown to me that 4o allotment has reset, and the result I get for the couple last prompt is from 4o, without any notification. I only realize that because of the notification that my 4o allotment has used up.

I am beginning to think that this is actually a feature, not a bug. The inability for the user to set default model to use in a conversation, like the removal of the menu option to set the model at the top of the screen, is thoroughly thoughtful and intentional, as the 4o Chat-GPT reply shows:

You as the user have no control on what the underlying model used

We are being Nudged out of 3.5.

If this is the case, I appreciate transparency from OpenAI about this, instead of backstabbing users.

1 Like

It doesn’t stay seted, the next response will always be in GPT-4, and changing manually is very tedious if the point is not to use up the GPT-4 accesses until necessary, especially if the uses are not restored even after you’ve used them up :confused: