Skywork Releases Open 'World Model' for Playable Video
The new 1.3-billion-parameter model functions as an interactive 'world model,' generating controllable video scenes from a single static image.

AI research lab Skywork has released Matrix-Game 2.0, a 1.3-billion-parameter model that generates interactive video from a single static image. Published with a permissive MIT license, the model is described as a "world model," designed to simulate a dynamic environment based on the content of an initial picture.
Unlike typical image-to-video generators that produce a fixed clip, Matrix-Game 2.0 is built for real-time interaction. According to the official release information, users can provide control inputs to direct the action within the generated video, effectively turning a static image into a playable, game-like scene.
This release provides an accessible, open-source tool for developers and researchers exploring simulated environments. By focusing on interactive control rather than just passive video generation, Skywork's model pushes further into the territory of creating dynamic, explorable digital worlds. Its relatively compact size makes it a notable entry in a field currently dominated by much larger, closed-source systems.
Sources
- Visit
Skywork/Matrix-Game-2.0
Hugging Face
More in Image → Video

MiniMax Releases H3 Video Model on Hugging Face
The company's new diffusion model handles text-to-video and image-to-video, with support for joint audio-video generation.

Wan-Dancer-14B turns still images into dance videos
Alibaba's Wan team releases an Apache-2.0 image-to-video model built for music-driven dance generation.

NVIDIA's Cosmos 3 Edge Brings World Models Closer
A new edge-optimized variant of NVIDIA's Cosmos world-model line aims to run generative video where the compute lives.
0 comments
No comments yet. Be the first to weigh in.