Cornac is a comparative framework for multimodal recommender systems. Cornac enables fast experiments and straightforward implementations of new models. It is highly compatible with existing machine learning libraries (e.g., TensorFlow, PyTorch).