MiniMax H3 is officially open source.

On August 3rd, the MiniMax H3 model was officially open-sourced. Domestic and foreign chip manufacturers such as Huawei Ascend, Moore Threads, Mu Xi, Haiguang Information, Kunlun Core, Tian Shu Zhixin, Biran Technology, AMD, Intel, as well as development communities and cloud inference platforms like Hugging Face, completed the relevant adaptation support on the same day. MiniMax H3 is a general-purpose multimodal generation system, ranking first in video editing capabilities on the Artificial Analysis leaderboard. It is capable of understanding multimodal contexts composed of text, images, videos, and audio, and can generate videos with a maximum resolution of 2K, lasting up to 15 seconds, accompanied by native stereo audio.