Comment by zozbot234

> good luck getting a machine that can run its specs though.

That's any machine that can physically host the weights and context. You'd need a highly-specced machine for better performance and throughput, but it's not a requirement as far as literally executing the model and getting output.