Changes to iOS app - model flipping

ChatGPT
1

Recent iOS App model changes have apparently removed the model selection switch for GPTPlus users. The problem is that conversations started with ChatGPT-4 will now switch models without notice to 3.5 mid-stream even though I will be nowhere near the query limit. So a conversation started in 4 starts producing significantly poorer quality responses, and there is no way to shift it back to 4.0. And of course it blatantly lies about what model is being employed. The only way to switch back to 4 is to start a whole new conversation, leading to a loss of prior context. This is a huge downgrade. Is anyone else seeing this? Note, it doesn’t seem to happen on the web interface, but on the iOS app it will switch models after only 1 or 2 follow-ups.

This is build 1.2023.347 (16603)

2

I have recently (past week) noticed this as well. I’ve asked chatGPT for it’s version, gotten the 4.0 response (and, this is indicated in the banner at the top) and then asked a substantive question and observe the banner switch to 3.5. I then ask chatGPT what version it is again. Sometimes it says 4.0, and then when pressed, says, 3.5. Sometimes it immediately says 3.5. This error is enough to kill the utility of the app. I will use the web interface for the time being, but would love to see someone fix this.