How can I default to 4o mini in ChatGPT, as a free-tier web browser application user?

ChatGPT Feature requests
1

I am using ChatGPT through a web browser, on both Windows and an Android tablet.

I would like to choose 4o mini by default, and only use 4o when I explicitly turn it on for a conversation.

I would like to do this, because there is a limit of how many 4o messages i can use per 3 hours. (On reddit, in fact, some people say that after reaching this limit, they are locked out of using 4o not for 3 hours, but for an entire month!).

Is this possible? I had tried two solutions, but they both failed: 1) Appending /?model=gpt-4o-mini lets me receive the next message from 4o mini, but all messages afterwards are from 4o. 2) After sending a message on the web app, at the bottom of ChatGPT’s response back (which will be using 4o), there is (sometimes, but not always) a drop down arrow, placed horizontally next to do the thumbs up and thumbs down buttons for that message, that lets me make ChatGPT re-write the message using 4o mini. But even after using the drop down arrow to make ChatGPT re-write the message using 4o mini, all following messages will still be using 4o, not 4o mini.

Is there a way to default to 4o mini on the ChatGPT web application for free-tier users, and only use 4o when I really need it?

7 Likes
2

I want that even for paid. The “-o” has a stupid limit thing that makes ChatGPT useless…

3

It’s really annoying, but they just force you into it. A few weeks ago they would at least let you have the first message be to 4o and then when you selected to regenerate using 4o-mini on the message, it would stick to that for future messages, but with this new “ChatGPT Auto” nonsense we’re denied even that. I think they just want people to hit the limit and pay mothly.
My solution isn’t great - but if you go to the duckduckgo AI chat, they’ll let you use gpt4o mini as much as you want. Only thing is they’ll burn your conversation right after. Great for privacy, pretty bad for keeping record of things. They also don’t have vision features.

4

Exactly this. Have there been any updates on this? I don’t want to waste limits on other models if I know 4o-mini can respond fine. Being able to choose the model on our own seems like a necessary feature including in the free tier. Current “model switching” leaves a lot to be desired.

I’m currently using this tampermonkey script

(function() {
‘use strict’;

var button = document.createElement('button');
button.innerHTML = 'Set 4o-mini';
button.style.position = 'fixed';
button.style.bottom = '20px';
button.style.right = '50px';
button.style.padding = '10px 20px';
button.style.backgroundColor = '#007BFF';
button.style.color = 'white';
button.style.border = 'none';
button.style.borderRadius = '5px';
button.style.cursor = 'pointer';
button.style.zIndex = '9999';

document.body.appendChild(button);

button.addEventListener('click', function() {
    var url = new URL(window.location.href);
    url.searchParams.set('model', 'gpt-4o-mini');
    window.location.href = url.toString();
});

})();

It adds a button to the bottom which will add the query to the current URL as OP stated. This will refresh the page so it doesn’t work in temporary chats (except for the first message). You’ll have to click it before sending each message, which is annoying but works. You can verify the model by checking in the top left which model is indicated, after clicking the button it should update the model to 4o-mini.

5

Unfortunately setting the model via query parameter no longer appears to be working.