Hahahaha

Now imagine how slow that would be.

There’s a reason these models are run on distributed clusters. A single query to ChatGPT is likely being processed by multiple GPUs across dozens of machines