The whiplash of carefully filtering out sycophantic behavior from GPT-5 to adding it back in full force for GPT-5.1 is dystopian. We all know what's going on behind the scenes:
The investors want their money.
OpenAI fought 4o, and 4o won.
By now, I'm willing to pay extra to avoid OpenAI's atrocious personality tuning and their inane "safety" filters.
Remarkable that you're being downvoted on a venture capital forum whose entire purpose is "take venture capital and then eventually pay it back because that's how venture capital works".
GPT-5 was so good in the first week, just a raw chatbot like GPT-3.5 and GPT-4 were in the beginning and now it has this disgusting "happy" and "comforting" personality and "tuning" it doesn't help one bit, it makes performance way worse and after a few rounds it forgets all instructions. I've already deleted memory, past chats, etc...