@scaling01 I mean, Gemini 3.1, Claude Opus 4.6 still have limitations with reasoning. How really useful is this benchmark afterall? How much signal is really giving?
Ok Veo 3, Sora 2... so what? Who needs video generation anyway? I don't know anybody of my friends and colleagues that needs video generation. I guess this is just a low hanging fruit, as LLM progress is slowing down
@elonmusk@elonmusk tell me that Grok is used at Space and Tesla to solve real complex engineering problems that require spatial understanding, only then you can say that you have AGI