China’s MiniMax curbs overseas access to new AI video model over copyright disputes

Chinese artificial intelligence firm MiniMax has released its new H3 video generation model as open-source, but has placed licensing restrictions on users in key overseas markets such as the United States and European Union. The decision comes amid ongoing copyright disputes and regulatory uncertainties surrounding generative video AI.

MiniMax released the model's weights, which encode its intelligence, to developers on Monday, allowing them to use, copy, modify, and distribute the software without paying a fee. However, developers in major markets are required to obtain a formal license to use the model, provided they meet compliance requirements and safeguards. Global access remains available through a paid subscription via MiniMax's website or authorized third parties.

The US restriction is specifically linked to ongoing generative video copyright litigation with major Hollywood studios, according to Ryan Lee, MiniMax's head of developer relations. The H3 model can generate videos up to 15 seconds long at 2K resolution, at a cost of 13 US cents per second. Despite Google's Gemini Omni Flash offering similar capabilities at a lower price of 1 US cent per second for 10-second clips, MiniMax's model is still seen as a strong competitor in the video editing space.

Licensing requirements for open-source video models have become a significant battleground in the generative AI industry, as Chinese developers strive to outpace US rivals by offering customizable and locally deployable models. MiniMax's H3 model has garnered attention for its advanced video generation capabilities and broad hardware compatibility.