Tencent’s large language model team Hunyuan has released an open-source image-to-video model, enabling users to generate five-second animated videos from a single image.

The model features lip-synced speech and singing animations generated from text or audio scripts, as well as motion-driven dance animations. It supports 2K high-quality output, with API access available through Tencent Cloud for enterprises and developers.

With 13 billion parameters, the model accommodates various styles, including realistic footage, animations, and CGI.

Hunyuan puts the tool open-source—including model weights, inference code, and LoRA training scripts—on GitHub and Hugging Face for customization.

Recent Tencent developments