MiniMax H3 brings text-to-video, image-to-video, reference-to-video, and video-to-video into one browser workspace. Create 4–15-second native 2K clips from prompts plus image, video, and audio references, with optional synchronized stereo sound. Multimodal references help guide identity, composition, motion, camera direction, timing, and style. The same workspace also includes text-to-image, image-to-image, and AI image editing with supported outputs up to 4K. It is built for creators, marketers, design teams, and independent makers who need fast visual iteration without switching between separate tools.