When we use any model is the processing happen locally on my system or on external server/gpu