Home / Technology / MiniMax Unveils H3: AI Video Generation Evolved
MiniMax Unveils H3: AI Video Generation Evolved
31 Jul
Summary
- H3 model processes text, image, video, and audio inputs.
- Generates 15-second, 2K resolution videos with stereo sound.
- MiniMax plans to release H3's model weights openly soon.
Chinese AI firm MiniMax has released its new H3 video-generation model, enhancing competition in the rapidly expanding AI market. This model uniquely processes text, images, video, and audio inputs to create sophisticated video content. It can generate videos lasting up to 15 seconds at a 2K resolution, complete with native stereo sound.
The H3 model offers advanced editing capabilities, including transferring movements between videos. MiniMax intends to make the model's weights publicly available within days, adopting an open-weight approach for video generation. This contrasts with many proprietary models currently dominating the field and allows users to download and customize the underlying system.
Founded in 2022, MiniMax is recognized as a significant player in China's AI startup scene. The H3 model is targeted towards commercial uses such as advertising, e-commerce, product design, and gaming. Notably, generating 2K video with H3 is projected to be less than one-third the cost of mainstream competitor products.
In alignment with China's drive to reduce reliance on U.S. semiconductors, MiniMax has designed H3 to function with various Chinese-made chips. This release intensifies the competition, following recent advancements from rivals like ByteDance with Seedance 2.0 and Kuaishou with Kling 3.0.
