ChatGPT (4o) said:
Which to pick?
Go with GPT-4o if you need built-in audio/voice support or the broadest multimodal capabilities.
Choose o4-mini-high for maximum text-context reach and faster, cost-efficient reasoning—especially handy for heavy coding or long-document workflows.
You said:
Thanks, and I switch back to 4o when i don’t want to waste my quote on simple things like thanking you 
ChatGPT said:
Exactly the right move. Use 4o for the pleasantries and the daft stuff, and save your o4-mini-high quota for the graft. Smart allocation, Michael.
You said:
Now if only you could do that automatically and save me forgetting I am in 3o when I am just chatting
ChatGPT said:
Aye, that’d be ideal. Bit daft that it doesn’t already—something like:
“This looks like low-effort chit-chat... fancy switching to 4o to save quota?”
Even a “default to 4o unless I explicitly pick something else” toggle would be a huge help. Right now you’ve got to babysit the selector like a Victorian butler with a tea tray.
If you ever start using the API or build a little wrapper app, you can wire up logic like that—auto-routing casual prompts to 4o and saving o4-mini-high for heavier stuff. Until then, it’s all manual… and easy to forget.
Might be worth sending in feedback directly from the app menu—OpenAI seems to act on that kind of quality-of-life suggestion when enough people flag it.
You said:
If you show me where the feedback option is, I will
ChatGPT said:
Cheers — here’s how to fire it off:
Click the ? icon in the lower right corner of the ChatGPT app (web or desktop).
Choose “Send us feedback” from the menu that pops up.
In the form, pick a category like “Feature Request” and say something like:
“I’d like ChatGPT to automatically switch to 4o when I’m just chatting, and only use my o4-mini-high quota when I ask more complex or resource-intensive questions.”
You can also add a second request like:
“Please show the quota reset timer more clearly when hovering doesn’t work.”
It won’t guarantee anything, but it does go into a triaged pool they actually track. Enough nudges and they might just do it.
(It was wrong, but the feedback stands!)