On August 3, MINIMAX-W (00100.HK) surged more than 10% on the Hong Kong Stock Exchange to HK$249.4, adding tens of billions of Hong Kong dollars to its market value in a single day. MiniMax unveiled its next-generation multimodal generation model H3 before the market opened, announcing it would open-source the model weights within days. The product, positioned as a "universal all-modal generation model," directly targets ByteDance's Seedance series and enters the market at 0.8 yuan per second. H3 supports unified understanding of text, images, video, and audio, and can output up to 15 seconds of 2K-resolution audiovisual content with native dual-channel audio. Its technical core lies in "Contextual Omni Representation," which treats multimodal inputs as a unified context. Its cross-modal composite reference capability is now on par with Seedance 2.5. In Artificial Analysis's video model benchmark, as of July 31, H3 ranked first globally in video editing, second in text-to-video, and third in image-to-video. According to Artificial Analysis official data, H3 ranks first globally in the video editing leaderboard with 1,130 Elo points, based on 5,043 blind preference samples, with statistics current as of July 31, 2026. On pricing, H3's API costs 0.8 yuan per second—one-twelfth of Seedance