Hola everyone,
Just reporting a few very specific bugs:
Android App TTS using Pro
When using the android app TTS—So, hitting the microphone for dictation and a TTS model is running over what is said then automatically sends the message for inference.
If you hit the ‘stop,’ there appears a little meta message below yours that says “message not sent.” This is not true if you are trying to use Pro. The TTS is converted and automatically sent. For how I understand pro to work, this is expected behavior.
So the bug is the TTS auto-send in the android app. This is unique to the Android, and not present in Chrome or the Windows Apps, which I think have a better solution where the TTS appends to the current message rather then auto sending; although I admit I found the feature convenient in mobile at first.
So, what happens is you send three messages to Pro instead of one and consequently use up the last of your carefully horded monthly pro credits. 
Wa-wa.
Windows App Long Conversations
While using the windows app and engaging in any number of very long conversations, I’ve noticed that the app starts progressively working more and more. The little spinner is constantly going up top left and there’s noticeable keyboard lag. Wa-wa.
It was pretty annoying till I realized it must be some sort of memory/summarization that’s running over my long topics every time.
So, especially for a native windows app, it’s probably cool to have some local memory and persistence! At least, I’d be good with that with some encryption; because having models summarize my long topics and going through various truncations and what not explains why models loose context from one turn to the next in long conversations if some long amount of time passes IRL.