So now that they’ve released 4o mini, as well as the Mac app which allows me to just open up ChatGPT as an app on 3 of my devices besides my windows pc, there’s a problem with how the models work and I think I’ve seen others from months ago and older ask as well. On iPad, I wanted to try and switch my default model to 4o mini, because it looks to be a better choice for me over 4o, and it’s unlimited use so why not just main it? However, if I select it, leave the app, and reopen it, it just goes back to 4o. There’s no option in the settings to change this which is super annoying, because you can have a whole memory, personalization, custom gpt’s that you can pin, but no way to select the default official model you want to use upon startup?
I have the same issue on the Web App (Windows).
There is no way to select what version of GPT you want for a specific chat.
By default it uses 4o but if you change the model it only re-generates that specific answer. The next question you ask in that chat will still use GPT 4o and not the mini version.
I would like an option in the chat settings that lets you select which version of GPT you want for the current chat, and every message to be generated using that version.
To chime in to this, it is ironic that in order to use the fast model, we have to wait for the slow model to finish first, and then generate the same prompt with the fast model. Kinda defeats the purpose of a fast model, at least for quick prototyping in a web browser.
Ah, an even worse downside I stumbled on: We have to quickly burn through all of our 4o quota, and then when we actually want to use 4o, we don’t have quota anymore. So I think this UX is actually causing OpenAI to lose money because it forces users to use 4o to the point of depleting their quota even when they don’t need it.
Yeah , it defeats the purpose. That’s why i’m gonna use copilot for less advanced questions instead of exhausting the quota of gpt 4o for them.
It feels like one of those strategies that corporations use to keep people from enjoying their free version, generally by either adding features which make it less helpful, or neglecting basic, necessary features that everyone expects to find, and is infuriated by the absence of.
Yet it doesn’t work, and just gets people to switch to other platforms. Why would OpenAI decide (and yes I am firmly of the opinion that this is deliberate - it lines up with their more recent decisions they have made to limit functionality, including adding a monthly limit on top of their daily limit for 4o usage) to sabotage their own platform?
Simple. Corporate myopia.
Such situations are where it is most effective to speak truth to power.
I am totally agreeing with you’ll. It’s super annoying to use 4o for me, because It’s slower than the earlier model I’ve been using. I don’t want to use model 4o all the time. I highly request the developer to chose the model and also add an option to only use 4o model when I’m telling to use internet or giving any file for the query.
Now that personalization settings exist, I would like the possibility to choose a default model, which is still not possible. Most of the times, for example, I am fine using gpt-4o mini, but I often end up using gpt-4o simply because I don’t notice it or I am lazy to switch. From your perspective, this means that I waste more energy using the top model without really needing that.
Agreed.
Defaulting to mini is annoying.
It’s kinda wild realizing that I’m blowing my o1 quota on pancake recipes instead of saving it for tough coding tasks. I do my best to use it sparingly, but I often open a new window, start typing, and submit, only to find out that I’m using my limited o1 quota. It’d be great to have more control over when I use it so I can savor it for the real heavy lifting!
Yea this extremely frustrating, o1 budget completely gone on Google replacement type searches when it can’t even connect to the internet
“I’m Also facing an issue with the ChatGPT app on my Windows laptop. When I try to set GPT-4 Turbo (4o mini) as the default model, it switches back to GPT-4 every time I reopen the app. There’s no option to permanently set the default model, which is really annoying. Despite features like memory and custom GPTs, it’s frustrating that this simple setting isn’t available. OpenAI should fix this as it’s causing a lot of problems.”