HunyuanVideo 1.5
Description
HunyuanVideo 1.5 is an open video generation model from Tencent. It has 8.3 billion parameters and creates video from text and from an image at 480p and 720p; there is a distilled variant that produces a clip in 8–12 steps, and a super-resolution module. The author opened not only the weights but also the training code and LoRA. Sound is not claimed in the author's model card. The weights are published on Hugging Face under the Tencent Hunyuan Community License. Per the license text, there are territorial restrictions, check the license, and companies with more than 100 million monthly active users must request a separate license from Tencent; other terms are in the license file in the author's GitHub repository. The minimum requirement per the author is 14 GB of VRAM with model offloading; on an RTX 4090 the distilled 480p image-to-video variant, per the author, produces a clip in about 75 seconds. The author's model card has no official demo without installation.
Pros and cons
Pros
- Compact 8.3B-parameter model, from 14 GB VRAM with offloading
- Text-to-video and image-to-video at 480p and 720p
- Distilled variant: a clip in 8–12 steps
- Training code and LoRA are open
- Super-resolution module included
Cons
- Per the license text, there are territorial restrictions, check the license
- Per the license text, more than 100 million monthly active users require a separate request to Tencent
- Sound is not claimed in the author's model card
- No official demo without installation
Key facts
- License
- Tencent Hunyuan Community License
- Commercial use
- Allowed with restrictions
- Hardware requirements
- At least 14 GB VRAM with model offloading (per the author)
- Developer
- Tencent
- Interface languages
- English
- Checked on
- Oct 17, 2026