Tencent releases HunyuanVideo inference code and model weights
Tencent released HunyuanVideo, a text-to-video model with more than 13 billion parameters, together with model weights and inference code.
What happened
Event details
Tencent released HunyuanVideo inference code and model weights on December 3, 2024. The text-to-video model had more than 13 billion parameters and used a dual-stream-to-single-stream diffusion Transformer, a causal 3D VAE, and a multimodal large language model as its text encoder. It could be deployed locally for video generation.
Assessment
Why it matters
HunyuanVideo brought an open-weight video-generation model with more than 13 billion parameters into the ecosystem and expanded the supply of large text-to-video models that could be deployed locally.
Availability
Access notes
Inference code and model weights were public under the Tencent Hunyuan Community License.