Model
Inference & Serving
HunyuanVideo
1Tencent's 2large 3video 4generation 5diffusion 6model
the six Ws · specification
W1
Who
Tencent released HunyuanVideo.
W2
What
HunyuanVideo is a large, open-weight text-to-video diffusion transformer model reported to outperform several closed commercial video generators on human evaluation benchmarks.
W3
Where
Hosted on GitHub at Tencent/HunyuanVideo and on Hugging Face under tencent.
W4
When
Released in December 2024.
W5
Why
It gave the open source community access to one of the largest publicly released video generation models at the time.
W6
With
Built on PyTorch with a causal 3D VAE and large multimodal text encoder, runnable via the official repo or ComfyUI.
W7
Watch
Released under the Tencent Hunyuan Community License, which permits broad use but restricts deployment for very large companies; weights are openly downloadable.