But if you followed them, there are focusing only on product for the last 2 years. The grand GPT-5 and their scaling laws, from which all their LLM AGI hopes originated, turned out to be a dud.