First of all, I believe customers should have access to other models. Claude, Llama 2, etc.

Next, I also believe that the limits should be a universal pool that any device can access (ex. one device is able to use 400,000 in one month, one uses 500,000, etc. vs. locking each device to 300,000– this is the only part of the deal that makes no sense to me whatsoever and may be the only reason I downgrade or refund entirely).

Lastly, GPT-4 Access should be considered as a user-enabled option, GPT-4 requests would use more tokens than GPT 3.5 requests. I have encountered MANY situations in which GPT-4 does a WAAAAY better job than GPT 3.5 prompted with the same request.

THE GOOD- It's easy to install and access. It's comparable to Viola, which I love and use all the time with the difference being that viola has optional web access, GPT 4 access, and does not impose limits on individual devices.

WISHLIST:

-Document Chat, similar to the AI training already available but would be nice to have the ability to base interactions on a PDF file for example.
-Standalone full screen mode, it would be nice to be able to access it as a window I can swipe to that only has the AI running
-Mobile app for IOS, I saw there was an android app but I believe more people are on iPhones–including myself
-AI Art limits possibly? entirely different thing but if a set number of tokens can be used to generate art that would be cool and make this a true all around AI solution
-Optional longer history retention, I saw in the description history is only saved for 2 weeks, there should be a local storage option that lasts indefinitely
-Document Editor Mode to use it somehow like a document analysis and editing software similar to prowritingaid and grammarly.