I’ve joined the forum specifically to request this.
Please provide a way to set a default model for Companion Chat AKA the Chat Bar on MacOS (I assume this applies to Windows as well).
Right now it seems to automatically use whatever model you last used in the main app, so if I was using the full app and an o1 model, next time I press [Option+Space] to open Companion Chat it will also use the o1 model.
This is frustrating because it eats into your quota over something you likely never would have purposely used that model for; Companion Chat is for quick info. Doubly frustrating is that you have no idea this is going to happen until it’s too late since there is no UI hint to inform you what model is being used (at least not until you’ve already submitted and see “thinking…” ). When you’re aware of this issue, you can never really use the Companion Chat without this worrisome thought in the back of your mind - not a good experience. Finally, I’m pretty sure the current behavior is just plain wasteful to the environment and more expensive for Open AI for the quick chat to be using o1 etc.
So again, please provide an option to set what model Companion Chat uses.
Hi,
I completely agree with your request and share the same frustration. Expanding on your point, this issue also applies to new chats in the macOS application, where the selected model can switch dynamically without notice. For instance, when I intend to use ChatGPT-4o, it often defaults to the o1-preview model, which quickly consumes the limited 50-message quota.
Having no way to control or set a default model creates unnecessary worry and diminishes the user experience, especially for quick interactions like Companion Chat. A clear option to define the default model would resolve this issue and provide consistency, ensuring users aren’t caught off guard and inadvertently using up limited resources. It would be a meaningful improvement across both the Companion Chat and the full app experience.
While they haven’t add a way to set a default, the latest version of MacOS app has made it very clear what model the companion chat is using and also gives us a way to change it. It also seems that the model selected is now decoupled between the companion and main apps - changing the model in main chat does not also change the model the companion chat is using - so as far as I’m concerned this issue has been addressed!
). When you’re aware of this issue, you can never really use the Companion Chat without this worrisome thought in the back of your mind - not a good experience. Finally, I’m pretty sure the current behavior is just plain wasteful to the environment and more expensive for Open AI for the quick chat to be using o1 etc.