Comment by _rs

That assumes you have the hardware capable of running inference on them at a reasonable speed