MiniMax has introduced a new multimodal generation model, MiniMax H3, capable of producing high-quality video clips with native stereo audio. This model processes various input formats, including text, images, video, and audio, as a unified context to generate 2K video clips lasting 4-15 seconds. The model is designed for general-purpose video generation, distinct from text-to-video models with additional features. The release of MiniMax H3 signifies advancements in video generation technology, which may have implications for various industries.