The platform provides a single interface for the three stages of world model research: training, evaluation, and control. It supports JEPA-based architectures and model-predictive control approaches using PyTorch. Designed for reproducible research, it includes standardized environments, solvers, and baselines for benchmarking world model performance.