World models · world · interactive · world_model
Matrix-Game 3.0
Matrix-Game
Matrix-Game 3 runtime is vendored under worldfoundry/synthesis/visual_generation/matrix_game/matrix_game_3_runtime.
Build your run
Choose a recorded variant, then copy the exact setup or inspection command.
Taskworld
Environmentworldfoundry-unified-cu128
DeviceCUDA 12.8
worldfoundry-eval evaluate \
--mode model \
--model-id matrix-game-3 \
--model-runner worldfoundry:pipeline \
--model-manifest-dir worldfoundry/data/models/catalog \
--requests-path tmp/requests.jsonl \
--output-dir tmp/model_eval/matrix-game-3 \
--metric artifact_count \
--jsonCompatibility & versions
Manifest-backed recipe
- Integration
- integrated
- Runner evidence
- verified
- Environment
- worldfoundry-unified-cu128
- Python
- 3.11
- CUDA
- CUDA 12.8
- PyTorch
- torch
- Source revision
- —
- Checkpoint revision
- 382aa382a03cec057360755aa72c7735025770bb
- Runtime profile
- matrix-game-3
- Pipeline binding
- matrix-game-3
- Runner
- worldfoundry.pipeline
- Pipeline target
- worldfoundry.pipelines.matrix_game.pipeline_matrix_game_3:MatrixGame3Pipeline
- Backend stage
- in_tree_runtime_plan
- Runtime status
- integrated
- Driver status
- compatible
- Environment kind
- Unified environment
Install environment
The environment resolver reads the recorded profile and chooses the unified or dedicated environment shown here.
bash scripts/setup/model_env_install.sh --model matrix-game-3Package constraints 18
torchtorchvisiontorchaudiotorchaoflash-attndiffuserstransformersacceleratesafetensorsimageioimageio-ffmpegnumpyscipypandastrimeshpilloweinopstqdm
Conda packages 4
python=3.11pipffmpegcuda-nvcc
Checkpoints & assets
Run the local check before allocating compute. Gated, private, and license fields below come directly from the checkpoint manifest.
worldfoundry-eval zoo model-download --model-id matrix-game-3 --check-local --jsonSkywork/Matrix-Game-3.0
- Revision
- 382aa382a03cec057360755aa72c7735025770bb
- License
- —
- Gated
- —
- Private
- —
Launch & outputs
The generated command uses the shared evaluation boundary and writes durable result manifests and artifacts.
worldfoundry-eval evaluate \
--mode model \
--model-id matrix-game-3 \
--model-runner worldfoundry:pipeline \
--model-manifest-dir worldfoundry/data/models/catalog \
--requests-path tmp/requests.jsonl \
--output-dir tmp/model_eval/matrix-game-3 \
--metric artifact_count \
--jsonInput contract
| Field | Recorded contract |
|---|---|
prompt | Required |
image | Required |
video | Optional |
actions | keyboard, mouse |
Artifact contract
| Artifact kind | Filename / path |
|---|---|
generated_world | matrix-game-3.mp4 |
Evidence & sources
Catalog integration, native-demo parity, and runner parity are independent records.
- Integration
- integrated
- Runner evidence
- verified
- Native demo evidence
- Not recorded
- Validation imports
- torch, torchvision, torchao, flash_attn, diffusers, transformers, imageio, numpy, scipy, pandas, trimesh, PIL, einops
Recipe provenance
worldfoundry/data/models/catalog/world_models/matrix-game-3.yamlGitHubhttps://github.com/SkyworkAI/Matrix-GameHugging Facehttps://huggingface.co/Skywork/Matrix-Game-3.0



