What is awesome-lfms-play-games?

Awesome-LFMs-Play-Games is a curated reading list of research papers focused on a specific question in AI: can large AI models, the same kind of models that power chatbots, learn to play games? The LFMs in the title stands for Large Foundation Models, which is an umbrella term covering language models (LLMs), vision-language models (VLMs), and other large neural network systems trained on broad datasets. The repository organizes papers into four categories: models (AI systems trained to play games), connecting frameworks (the infrastructure used to link AI to game environments), datasets (collections of gameplay data used for training or evaluation), and benchmarks (standardized tests that measure how well an AI plays). The focus is mostly on research from 2025 onwards, with a small number of classic earlier works included for context. This kind of research matters because games offer controlled, measurable environments for testing whether AI can make complex decisions, plan ahead, adapt to changing situations, and cooperate or compete with other agents. Games covered span board games like chess and poker, video games like Minecraft, and general simulation environments. You would consult this repository if you are an AI researcher, student, or enthusiast following the frontier of AI agents and game-playing, and want a single organized collection of relevant papers rather than searching the academic literature yourself. There is no runnable code, it is a reference list maintained as a public resource. The full README is longer than what was shown.