[properbrew]

+1 for OpenVINO, we utilise it for our model. It's quite amazing the inference speed you can get from CPUs that most people would assume are running on a GPU.