I can't think of many places to share this, so I might as well share with you guys, since this is also about genAI and (tangentially) related.

A while ago, while ChatGPT was 4o / 4.5, I discovered that you could endlessly abuse the message limit by cycling through 3 VPNs in 3 different countries. Literally just go something like USA->France->Germany->USA (while loading the website once per connection) and you have your 4o generations again instead of mini. Without ever logging out of your user account.

Specifically, it seemed like this was creating some kind of "instances" (idk; for lack of a better word) or "new unique sessions", which of course shared the message history between each other, but seemed to be otherwise separate despite belonging to the same user. Once, I saw that a new chat created "from one country" was not available in the chat list on the left when viewed from other countries, but only once. It was still available through the direct URL.

Now, with GPT-5, I see the same behavior, but now it jumps between GPT-5 and GPT-4o. I'm a Free user and I have 4o because of this. Seriously. It's not that it's enabled as a fallback after free gens of 5 (because 5 goes to 5-mini for Free users); it's specifically that in some of the sessions ChatGPT thinks that I'm supposed to be on 4o. 4o gets 4.1-mini as a fallback when I'm out of gens (just like it did in the past), and attempts to "Try again" on previous replies result in the error "Model not found" (if I am "currently on 4o" and doing "try again" on a GPT-5 response, or vice versa).

Seems fun at a glance, but the models behave and write differently, and it's erratic and random and annoying. I can have GPT-5 on my phone and 4o on my PC right at the same time, when logged into one and the same account, and I see no way to "make it" switch to the different model (regardless of whether or not I'm out of free gens on either of them).

It's all very strange and idk how to even report this obvious bug.