MIRA: A Fully Hallucinated Multiplayer Rocket League Powered by World Models
mathemagic1an · x · 2026-07-29
General Intuition and Kyutai open-sourced MIRA, a video world model that simulates Rocket League entirely through generation. Without any physics or rendering engines, it uses a 5B-parameter diffusion transformer and a 600M-param codec to enable 4-player real-time browser multiplayer at 20fps.
Purely learning from data, the model accurately simulates complex game mechanics: moving, kicking, tracking boost gauges, and demolitions. By using action dropout during training, the model also features an autopilot mode. The dataset and code are fully open-sourced.
More from Multimodal
- AI video turns a kidnapping premise into a self-duplication joke — umesh_ai · 2026-07-29
- Sonder Editor brings an open-source video timeline to ComfyUI — SonderSaid · 2026-07-29
- Claude gaming art leaps: year-over-year comparison is stunning — iamfakhrealam · 2026-07-29
- SDXL Image Gen: How to Build Complex Multi-Character POV Interactions — ZeHirMan · 2026-07-29
- Hailuo AI’s new video model reportedly handles up to 12 image, video, and audio references — aziz4ai · 2026-07-29
- PDD accelerates image and video diffusion by predicting multiple denoising steps at once — ArashVahdat · 2026-07-29