The process of configuring and launching a trained model in a cloud environment so it can receive requests and generate responses.