What it does
MiniMax presents itself as a company building multimodal models, and this page is the front door to its model families and apps. The language line includes MiniMax M3, a coding and agent model with a 1M context and native multimodality, alongside M2.7 and M2.5. Video generation is covered by MiniMax H3, and the speech and music line lists Speech 2.8 and Music 3.0. On the product side the site lists MiniMax Code, a coding agent made to fit MiniMax models, with separate Coding and Work modes, and MiniMax Design, an agent-led platform for making advertising, e-commerce, brand and creative post-production content. Its navigation also names a voice product and an app called Xingye. Developers are pointed to an open platform for businesses and developers, with guides for text generation, video generation and voice cloning, and the site shows subscription entry points for its apps. The research section lists recent write-ups, including an open-weights release of Music 3.0. The page text is in Chinese, with an English version available.
Capabilities
- M3 language model with a 1M context and native multimodality
- H3 video model, plus Speech 2.8 and Music 3.0
- MiniMax Code agent with separate Coding and Work modes
- MiniMax Design for advertising, e-commerce and brand content
- API access to the models, with guides for developers
When to use it
- You want to see MiniMax's current text, video, speech and music models in one place
- You are looking for a coding agent tuned to MiniMax models
- You produce marketing or e-commerce content and want an agent-led workflow
- You need the entry points for MiniMax API guides and subscriptions
