Grok has capabilities in understanding our physical world. Grok outperforms its peers in our new RealWorldQA benchmark that measures real-world spatial understanding. For all datasets below, they evaluate Grok in a zero-shot setting without chain-of-thought prompting. This leading real world understanding is building upon Tesla data and work with FSD (full self driving).

Recent reports are that Tesla has 100,000 Nvidia H100 chips or more. This would be 400 exaflops. Elon has said that XAI needs to train Grok 3 with 100,000 Nvidia H100 chips. It is clear that XAI will push future versions of Grok for real world AI leadership in understanding video and audio.