World models · world · video · world_model

HY-WorldPlay

Tencent Hunyuan / HY-World

Existing in-tree runtime is now exposed through a BaseSynthesis subclass.

Runner verifiedUnified environmenthy-worldplay

Build your run

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

worldfoundry.pipeline

Variant

Taskworld

Environmentworldfoundry-unified-cu128

DeviceCUDA 12.8

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

Compatibility & 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: f4c29235647707b571479a69b569e4166f9f5bf8

Runtime profile: hy-worldplay
Pipeline binding: hy-worldplay
Runner: worldfoundry.pipeline
Pipeline target: worldfoundry.pipelines.hunyuan_world.pipeline_hunyuan_worldplay:HunyuanWorldPlayPipeline
Backend stage: official_in_tree_runtime
Runtime status: official_8gpu_studio_demo_verified
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 hy-worldplay

Package constraints 8

torch
torchvision
numpy
opencv-python
diffusers
transformers
loguru
decord

Conda packages 3

python
pip
ffmpeg

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 hy-worldplay --check-local --json

tencent/HY-WorldPlay

Revision: —
License: —
Gated: —
Private: —

tencent/HunyuanVideo-1.5

Revision: —
License: —
Gated: —
Private: —

Qwen/Qwen2.5-VL-7B-Instruct

Revision: —
License: —
Gated: —
Private: —

google/byt5-small

Revision: —
License: —
Gated: —
Private: —

black-forest-labs/FLUX.1-Redux-dev

Revision: —
License: —
Gated: —
Private: —

tencent/HY-WorldPlay

Revision: f4c29235647707b571479a69b569e4166f9f5bf8
License: —
Gated: —
Private: —

tencent/HunyuanVideo-1.5

Revision: 9b49404b3f5df2a8f0b31df27a0c7ab872e7b038
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 hy-worldplay \
  --model-runner worldfoundry:pipeline \
  --model-manifest-dir worldfoundry/data/models/catalog \
  --requests-path tmp/requests.jsonl \
  --output-dir tmp/model_eval/hy-worldplay \
  --metric artifact_count \
  --json

Input contract

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

Artifact contract

Artifact kind	Filename / path
`generated_world`	`hy-worldplay.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, numpy, cv2, decord, loguru

Recipe provenance

worldfoundry/data/models/catalog/world_models/hy-worldplay.yamlGitHubhttps://github.com/Tencent-Hunyuan/HY-WorldPlayHugging Facehttps://huggingface.co/tencent/HY-WorldPlay

Build your run

Input contract

Artifact contract

Related recipes