HunyuanVideo is an open-source generative video model family developed by Tencent that creates videos from textual and visual inputs using a unified Transformer architecture with flow matching. The model supports text-to-video, image-to-video, and avatar animation tasks, utilizing a 3D Variational Autoencoder for spatiotemporal compression and a multimodal language model for enhanced text-visual alignment across various resolutions and video lengths.