HunyuanVideo-1.5
HunyuanVideo-1.5 combines an 8.3B diffusion transformer with a causal 3D VAE. Its released checkpoints support text-to-video and image-to-video generation.
Model design and access
The repository includes 480p distilled variants, selected 720p variants and super-resolution models. A 480p image-to-video step-distilled checkpoint recommends eight or twelve steps. Some planned variants remain marked as forthcoming; they should not be treated as released weights.HunyuanVideo-1.5 official model documentation
Practical use
Practical guidance. Select the exact task and resolution checkpoint before measuring speed. A super-resolution stage changes the final file size without proving that all detail was generated natively at that size. Review motion and identity consistency across the whole clip.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| License | Tencent Hunyuan Community License; custom terms applyHunyuanVideo-1.5 official model documentation |
| Native variants | 480p and 720p; task-specific checkpointsHunyuanVideo-1.5 official model documentation |
| Documented minimum | 14 GB GPU memory with model offloadingHunyuanVideo-1.5 official model documentation |
Related models
- HunyuanVideo
Tencent · HunyuanVideo
HunyuanVideo is the original text-to-video release published in December 2024.
Launch date Downloadable - HunyuanVideo-I2V
Tencent · HunyuanVideo
HunyuanVideo-I2V extends the original HunyuanVideo model with reference-image conditioning.
Launch date not confirmedDownloadable