I wonder how fast it performs on just a CPU? If the model performs say 10x on a GPU cluster, would it also perform faster on a CPU? This could bring proper des

I wonder how fast it performs on just a CPU? If the model performs say 10x on a GPU cluster, would it also perform faster on a CPU? This could bring proper desktop AI to the average laptop user, which could be a game changer for running local models.