Absolutely. Even in extended thinking mode it was thinking for only a few seconds in prompts that used to take minutes. Much faster token/s in any mode and significantly worse, exactly as you describe.

It seems like they might still be heavily nerfing / quantizing the models in production a couple weeks before a new release, like they have always (unofficially) done.

I did notice that, I thought maybe I’d exceeded my thinking requests

GPT-5 was horrible. It produced AI slop have immense speed, which is quite tough when other coworkers ask to review their PR...