The time or latency it takes for a machine learning model to produce results from input data.