While general vibe of Sam and OpenAI losing steam is correct for me; I genially disagree with all “hate” of GPT-5. It was indeed not a crazy breakthrough, but, honestly, 5.2 model at extended thinking path in Pro version is utterly scary to me. I can give it some complicated multi-tier question which requires modeling, abstract thinking, computation, and rationality; it can walk away for 45 minutes, produce CoT with a length of Empire State Building and give me fairly good and well written response. I can’t speak for all industries, but 5.2-Pro-extended is really scarce when it comes to math and reasoning. Here I am a bit on Sam side when he said that most of people severely underutilize modern AI by using it same as in 2023. The capability of recent models sounds to me way beyond current typical use cases.