HunyuanVideo
TencentHunyuanVideo is Tencent's family of diffusion-transformer video generation models. The original 13B text-to-video model and its image-to-video variant were followed by HunyuanVideo 1.5, an 8.3B model with a 3D causal VAE that generates at 480p and 720p and upscales to 1080p with a few-step super-resolution network. Tencent publishes the weights with inference code and, for 1.5, a full training pipeline.
Openness
3 high confidence- weights
- open(HunyuanVideo, HunyuanVideo-1.5 and HunyuanVideo-I2V download from Hugging Face)
- data
- described(the READMEs describe data curation
- code
- open(HunyuanVideo-1.5 publishes train.py, a full distributed training pipeline)
- license
- Tencent-Hunyuan-Community-License(every checkpoint: licensed for a territory that excludes the European Union, the United Kingdom and South Korea
Tencent's community license allows commercial use but does not apply at all in the European Union, the United Kingdom or South Korea, and requires a separate license for the very largest services. HunyuanVideo 1.5 publishes its full training pipeline, while the training data is described but not released. The Tencent Hunyuan Community License bounds commercial use and excludes the EU, UK and South Korea, and a territorial exclusion caps a license at use_bounded.
- https://huggingface.co/api/models?author=tencent&sort=downloads&direction=-1&limit=100&search=HunyuanVideo recorded 2026-09-26
Hugging Face listing of HunyuanVideo repositories: HunyuanVideo, HunyuanVideo-1.5 and HunyuanVideo-I2V public, license tencent-hunyuan-community.
- https://raw.githubusercontent.com/Tencent-Hunyuan/HunyuanVideo-1.5/main/LICENSE recorded 2026-09-26
License body: "TENCENT HUNYUAN COMMUNITY LICENSE AGREEMENT"; "THIS LICENSE AGREEMENT DOES NOT APPLY IN THE EUROPEAN UNION, UNITED KINGDOM AND SOUTH KOREA"; licensees "greater than 100 million monthly active users ... must request a license from Tencent".
- https://raw.githubusercontent.com/Tencent-Hunyuan/HunyuanVideo-1.5/main/README.md recorded 2026-09-26
README: "We now open-source the training code for HunyuanVideo-1.5! The training script (train.py) provides a full training pipeline"; the model is credited to "meticulous data curation", with no dataset released.
Adoption
1 high confidenceAdoption is measured as Hugging Face downloads across the three HunyuanVideo checkpoints. Converted copies on other accounts are not counted.
- https://huggingface.co/api/models?author=tencent&sort=downloads&direction=-1&limit=100&search=HunyuanVideo recorded 2026-09-26
Listing sorted by downloads: HunyuanVideo 1269, HunyuanVideo-1.5 1193, HunyuanVideo-I2V 179 in the trailing 30 days.
Capability
2 medium confidenceHunyuanVideo 1.5 places in the lower half of the video arena, a step below LTX-2.5 and below the downloadable Wan 2.2. MiniMax H3 leads the downloadable video models.
- https://artificialanalysis.ai/video/leaderboard/text-to-video recorded 2026-09-26
Leaderboard entries "HunyuanVideo-1.5 (Fal)" (rank 67, Elo 1020.0, open weights at huggingface.co/tencent/HunyuanVideo-1.5) and "Hunyuan Video (Fal)" (rank 74).
Verified 2026-09-26