HunyuanVideo
HunyuanVideo is the original text-to-video release published in December 2024. It remains available with model weights and sampling code.
Model design and access
Its design combines image-video training, a multimodal language model for prompt encoding and a 3D variational autoencoder. The repository includes a 720p workflow, distributed inference and an FP8 weight format. HunyuanVideo-I2V and HunyuanVideo-1.5 are separate releases.HunyuanVideo official model documentation
Practical use
Practical guidance. Retain this model for reproducible work based on the original pipeline. If starting a new project, compare whole-clip motion and hardware use with 1.5 rather than assuming a newer release has the same visual behavior.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Launch date | |
| Access | Downloadable |
| License | Tencent Hunyuan Community License; custom terms applyHunyuanVideo official model documentation |
Related models
- HunyuanVideo-1.5
Tencent · HunyuanVideo
A text-to-video and image-to-video model combining an 8.3B diffusion transformer with a causal 3D VAE.
Launch date not confirmedDownloadable - HunyuanVideo-I2V
Tencent · HunyuanVideo
HunyuanVideo-I2V extends the original HunyuanVideo model with reference-image conditioning.
Launch date not confirmedDownloadable