Advances in natural language processing are largely based on research using statistical regularities of fixed language data, without an explicit understanding of the real-world. We built LIGHT (Learning in Interactive Games with Humans and Text), a large-scale fantasy text adventure game research platform for training agents that can both talk and act, interacting either with other models or with humans. The game uses natural language that’s entirely written by people playing the game. This platform enables researchers to study language and actions jointly in a game world. We've made the complete setup open-source and available to other researchers.