In Auto mode it relied on ChatGPT-5 (was familiar with its operation, guessed later), 14 days trial lasted exactly for 3 hours. Nice. Good business strategy.
AUTO mode is not GPT-5, and it’s completely free.
Just look AUTO mode is very fast and doesn’t think at all. How could you think it’s GPT-5?
Well, I am maybe wrong. I was slow as hell. Its replies reminded me to GPT5. But if you identified THIS as a problem and not that a 14 trial worked for 3 hours….
Did the model think or not?
And at which stage was it slow?
Before it even loaded, or did the response arrive slowly, token by token?
It was slow, thinking, halting, AFAIK even ‘thinking loudly’. Still do not care: it was 3 hours. And by default it installed itsekf with GPT5 as default agent; i had to switch to auto when I discovered as I have terrible experience with GPT5 on Copilot.
I understand that for some reason you had an unpleasant experience, and that’s reasonable given the system and the code you’re working on, which might not be that complex.
If you don’t need it, you don’t have to use it but there are things it’s amazing at.
Your experience is very personal.
Personal experiences result personal decisions. Many bad personal experiences result - overall - collective decisions.
None of these change that my trial lasted for 3 hours.