A model that has already been trained on a data set for a specific machine learning task. You can use pretrained models as is, or use transfer learning to customize the model for a specific task. Also called pre-trained networks if they’re pre-trained neural network models. Pre-trained models are ML models trained on a large data set to solve problems similar to the one that we want to solve. Due to the computational cost of training such models, it is common practice to import and use models that others have already trained, some free and open source.