World models · interactive-world-model · image-to-video · camera-control

Hunyuan Game Craft

tencent

In-tree Hunyuan GameCraft pipeline is registered; official inference uses the HYGameCraft conda profile and torchrun dispatch.

Runner verifiedDedicated environmenthunyuan-game-craft

Build your run

Choose a recorded variant, then copy the exact setup or inspection command.

worldfoundry.pipeline

Variant

Taskinteractive-world-model

EnvironmentHYGameCraft

DeviceCUDA 12.4

worldfoundry-eval evaluate \
  --mode model \
  --model-id hunyuan-game-craft \
  --model-runner worldfoundry:pipeline \
  --model-manifest-dir worldfoundry/data/models/catalog \
  --requests-path tmp/requests.jsonl \
  --output-dir tmp/model_eval/hunyuan-game-craft \
  --metric artifact_count \
  --json

Compatibility & versions

Manifest-backed recipe

Integration: runtime ported
Runner evidence: verified
Environment: HYGameCraft
Python: 3.10
CUDA: CUDA 12.4
PyTorch: torch==2.5.1
Source revision: —
Checkpoint revision: 022caa8698ea7134cd7823875757e7322fe894d6

Runtime profile: hunyuan-game-craft
Pipeline binding: hunyuan-game-craft
Runner: worldfoundry.pipeline
Pipeline target: worldfoundry.pipelines.hunyuan_world.pipeline_hunyuan_game_craft:HunyuanGameCraftPipeline
Backend stage: official_in_tree_runtime
Runtime status: official_torchrun_contract
Driver status: compatible
Environment kind: Dedicated 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 hunyuan-game-craft

Package constraints 27

torch==2.5.1
torchvision==0.20.0
torchaudio==2.5.1
accelerate==1.9.0
av==15.0.0
decord==0.6.0
diffusers==0.34.0
einops==0.8.1
huggingface-hub==0.34.3
imageio==2.37.0
imageio-ffmpeg==0.6.0
loguru==0.7.3
numpy==2.1.2
nvidia-ml-py==12.575.51
nvitop==1.5.2
opencv-python-headless==4.12.0.88
packaging==25.0
pillow==11.0.0
protobuf==6.31.1
psutil==7.0.0
pyyaml==6.0.2
safetensors==0.5.3
sentencepiece==0.2.0
tokenizers==0.21.4
tqdm==4.67.1
transformers==4.54.1
triton==3.1.0

Conda packages 5

python=3.10
pip
ffmpeg
ninja
cuda-toolkit=12.4

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 hunyuan-game-craft --check-local --json

tencent/Hunyuan-GameCraft-1.0

Revision: 022caa8698ea7134cd7823875757e7322fe894d6
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 hunyuan-game-craft \
  --model-runner worldfoundry:pipeline \
  --model-manifest-dir worldfoundry/data/models/catalog \
  --requests-path tmp/requests.jsonl \
  --output-dir tmp/model_eval/hunyuan-game-craft \
  --metric artifact_count \
  --json

Input contract

Field	Recorded contract
`prompt`	Required
`image`	Required
`video`	Optional
`actions`	camera_path, navigation_action

Artifact contract

Artifact kind	Filename / path
`generated_world`	`hunyuan-game-craft.mp4`

Evidence & sources

Catalog integration, native-demo parity, and runner parity are independent records.

Integration: runtime ported
Runner evidence: verified
Native demo evidence: Not recorded
Validation imports: torch, torchvision, cv2, imageio, loguru, flash_attn

Recipe provenance

worldfoundry/data/models/catalog/world_models/hunyuan-game-craft.yamlGitHubhttps://github.com/Tencent-Hunyuan/Hunyuan-GameCraft-1.0Hugging Facehttps://huggingface.co/tencent/Hunyuan-GameCraft-1.0

Build your run

Input contract

Artifact contract

Related recipes