MiniMax announced Hub, a multimodal AI video generator that integrates image creation, video, narration, music, and editing into one platform. Xu Lüyang from MiniMax’s Product Operations department introduced the tool at the Shanghai International Film Festival.
“Before, we would open one AI tool for images, another for video, then another for voiceover or music, and then finally use a video editor to stitch everything together,” Xu says. “Now, you simply tell Hub your goals in natural language or drop a PDF proposal, reference video, or asset pack. The AI agent within Hub automatically understands your requirements, breaks down tasks, selects a model, runs it, and validates quality. We simply step in to review and adjust at key points. It’s not just a smarter ‘generate’ button, it’s a creative partner that knows how to execute your project. ”
Central to the Hub’s design is a human-involved approach, with the tool pausing at key decision points rather than acting as a one-click generator. “AI agents should not be black box, one-click generators, pausing at every critical decision point for confirmation. We believe that AI should take the burden of execution off humans’ shoulders, but creative direction and aesthetic judgment should ultimately rest with humans,” said Xu.
The Hub also includes “Skills” and “Memory” features that adapt the tool to individual users. “We hand over workflows, aesthetic standards, and prompt engineering expertise to agents, and let them remember them so they can perform the task independently next time,” says Xu.
Pan Yuying, head of video content at MiniMax Multimodal, demonstrated additional features such as document generation across multiple file formats.
MiniMax is collaborating with AI Backlot, an AI work lab in Shanghai, and four pairs of creators from this year’s group are using the Hub to create short films.
Shanghai-based MiniMax develops multimodal AI models and applications, including AI character apps Talkie and Xingye and video generator Hailuo AI.
