In the past, the injected OpenAI preprompt was different on the mobile app vs the browser. I’m not sure if that is still the case? It’s kinda hard to find out unless you “jailbreak” it.
No clue if there is an app to download or anything, but, I wrote my own in C++. The advantages I see are, well, it’s not web based. No tracking, etc. It’s very fast, since it isn’t or does not use Python.
I kept it text based, but, it’s really easy to add images, audio or video. It was built for a music application, but, I decided to also use it as an assistant.
Since it’s API based, I had to code the memory for it (how it can keep a conversation going). I am able to generate multi part music compositions with it using a que, too. (midi based, not, audio)
Openai has a C++ library that is used, has all the examples you would need too.
I predominantly use the Desktop app on my Mac (see here). I do find the desktop app to be much more snappy!
I also use the app on my iPhone, but as others have pointed out, mostly in voice mode. On the phone it’s come in very handy where I needed to take a picture and get some insights (e.g. I used it to “diagnose” one of my plants). The iPhone app is also very snappy.
Yeah +1 on dev tools - when I want to troubleshoot something, I switch over to browser based ChatGPT (Chrome), do the inspect, check the network requests and responses.