Researchers from General Intuition, Kyutai, and Epic Games released MIRA, a 5B parameter interactive world model trained on 10k hours of synthetic Rocket League data, enabling real-time 4-player gameplay at 20 FPS on a single B200 GPU. MIRA demonstrates the feasibility of large-scale world models for multiplayer, physically complex environments, potentially accelerating research in game AI, simulation, and reinforcement learning by providing a playable demo and open-source tools. The model uses a latent diffusion architecture to generate video frames conditioned on all four players' actions, and the team released a 1,000-hour dataset of 4-player gameplay, a technical report, and a playable online demo.
Background
World models are neural networks that learn to simulate the dynamics of an environment, enabling agents to plan and reason. Previous single-player world models treat other agents as part of the environment, but MIRA conditions on multiple agent actions, making it suitable for multiplayer games. Rocket League is a high-speed physics-based game where players control rocket-powered cars to hit a ball into goals.