All the computations will be run using a CPU. A GPU-based model can be deployed but the inference will be processed on a CPU.