Tencent launches Hunyuan, a new open-source AI video model competing with Runway

Tencent has launched Hunyuan, a new open-source AI video model that competes with established players like Runway. Hunyuan can generate high-resolution 5-second videos from text prompts and features 13 billion parameters, making it one of the largest models available. Currently, Hunyuan is primarily accessible in China, but its open-source nature allows for broader experimentation. A demo on FAL.ai showcases its ability to create photorealistic motion, although initial tests indicate it may not match the prompt adherence of competitors. While Hunyuan shows promise, it requires significant GPU resources, limiting its accessibility. Early evaluations suggest it may need further refinement to improve its contextual understanding and output quality compared to other models.