GPT4o was FINALLY starting to work well without me having to repeat the tasks 30x over before it gets it right without errors or fading out in the quality of the response as it goes along, and it would remember conversations and work really good - and they suddenly decided in end of April to mess it up completely, replace it, then reinstated it to some prior version that was awful and got back to square one with errors.
And now we have 4o and 4.1 with constant reprompts and errors, poor memory, not able to complete tasks without redoing them, etc…
WHO makes these decisions???