MiniMax Design

MiniMax Design is MiniMax's own desktop multimodal creation agent: describe a concept or drop in a brief, and a lead agent decomposes the task, assigns specialist agents, picks the right models across text, image, video, voice, music and editing, generates assets and manages the workflow with review and export. It runs as a native macOS/Windows app (M-series and Intel supported), with a visual canvas, reusable Skills, local asset indexing and professional tool integrations. Chinese-market native, with a Feishu community.

A first-party model lab shipping its own agent studio (rather than licensing models to wrappers) is the notable structural move — the stack from model to workflow is theirs end to end.

Who it's for: Chinese-speaking creators and teams who want MiniMax model quality with agent orchestration on desktop, and anyone evaluating how first-party labs are productizing "agent does the whole creative pipeline." If you're outside the Chinese ecosystem, the Feishu-centric community and localization are the main friction points — the tool itself is real.