Shafaqna English– MiniMax, a Chinese AI developer, introduced its latest video-generation model on Friday(31 Jul 2026), which supports multimodal inputs spanning text, images, video, and audio. The release marks a significant push by the company into a fast-growing segment of the AI market, where major players like ByteDance and Kuaishou are already well-established leaders.
According to the Shanghai-based developer, the H3 model can create videos of up to 15 seconds at 2K quality, complete with integrated stereo sound. In addition, it offers users the ability to edit existing footage and to transfer motion characteristics from one video to another, using simple commands and reference assets supplied in various formats to achieve the desired results.