Research framework that turns the Minimax H3 video generator into an interactive world model controlled by key presses.
Updated EU status checked Last shown
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
H3-World is a research project that adapts the Minimax H3 video generation model into an interactive, game-like world model. Key presses are translated into short text descriptions, which are fed to the model's language layer to steer character and camera movement in the generated video.
According to the project page, it enables character and camera control across diverse visual environments. A YouTube review reports that code and a training script have been released, though no repository link was available in the material provided, so details such as license and weights could not be verified.
It is aimed at researchers and developers exploring controllable video generation and real-time world simulation. Practical limits, such as latency, resolution, and hardware requirements, are not stated in the available material.
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Turns the open-source Minimax H3 video generator into an interactive video game engine by translating key presses into short text descriptions fed to the model's language layer. Code and training script are released.
Closest neighbours by meaning, not just by category.
Research
Research framework where a coding agent maintains world state as code, rendered to video by a fine-tuned MiniMax-H3 model.
Model · MiniMax
Open-weight video generation model from MiniMax, part of its H3 family alongside H3 Image.
Model
Real-time interactive world model with compressed memory and hybrid action and semantic control, built on Wan 2.2.
Model · NVIDIA
NVIDIA research system where a code game engine drives a two-step streaming MiniMax-H3 video model for real-time rendering.
Model
Interactive autoregressive world model generating minute-long playable video with real-time camera control and prompt events.
Model · Adobe
Adobe Research video world model that builds navigable interactive worlds from an image or video at 16 FPS.