# SkyworkAI/Matrix-Game

Matrix-Game 3.0: Real-Time and Streaming Interactive World Model with Long-Horizon Memory

Repository: https://github.com/SkyworkAI/Matrix-Game
Canonical: https://ross.abutalabs.com/products/matrix-game
Homepage: https://matrix-game-v3.github.io/
Language: Python
License: MIT
License Family: permissive
Topics: genie, interactive-video, long-sequence, long-video, real-time, video-generation, world-model, long-horizon-memory
Last push: 2026-03-30T06:32:00+00:00

## Health v2 (maintenance only)
Score: 52/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 74, release rhythm 35, longevity 34
- inputs: {"age_days": 484, "days_push": 156, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2314, forks 250 (observed 2026-08-28T04:06:36.733860+00:00)

## What it is
Matrix-Game is Skywork AI's open-source series of interactive world foundation models that generate real-time, streaming video in response to user actions. Version 3.0 adds long-horizon memory for temporally consistent 720p long-form interactive video generation.

## Use cases
- generate interactive video from keyboard or gamepad actions
- build a playable neural world model like Genie
- stream long videos in real time with a diffusion world model
- research memory-augmented video generation for temporal consistency
- create synthetic game environments from video-pose-action data
- distill diffusion models for real-time inference

## When to choose
- you need an open-source interactive world model with real-time streaming generation
- you are researching long-horizon consistency or memory in video generation
- you want MIT-licensed world model weights and official implementations

## When to avoid
- you need a production game engine rather than a research model
- you lack high-end GPUs for real-time diffusion inference
- you need fine-grained control over a specific existing game world

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, video-processing, llm-inference, simulation
- domain: artificial-intelligence, deep-learning, simulation
- platform: python, cross-platform
- tags: world-model, interactive-video-generation, real-time-video, diffusion-models, long-horizon-memory, video-generation, genie, streaming-generation, model-weights, video, game-development, gpu, linux

## Member repositories
- SkyworkAI/Matrix-Game (main) score 52

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:36.733860+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T02:39:26.628107+00:00, confidence not recorded.
  - readme: https://github.com/SkyworkAI/Matrix-Game (fetched 2026-08-28T04:06:36.733860+00:00, sha 1a0b543f1e00)
  - homepage: https://matrix-game-v3.github.io/ (fetched 2026-08-29T10:19:23.300704+00:00, sha 9c3b277871ff)
- Data as of 2026-08-30T08:39:29.467469+00:00.