MiniMax Releases MiniMax-Music3: An Open-Weights Music Model Generating Complete Five-Minute Songs From Lyrics and a Structured Caption

MiniMax has unveiled MiniMax-Music3, a groundbreaking open-weights text-to-music model that represents a significant advancement in AI-driven music generation. The model can generate complete songs up to five minutes in length from user-provided lyrics and structured captions in a single processing pass, outputting high-quality 32 kHz, 16-bit stereo WAV files. This development marks an important milestone in democratizing music creation technology by making the model publicly available to developers and researchers worldwide.

MiniMax-Music3 employs a sophisticated architecture designed to handle long-form music generation efficiently. The model accepts two primary inputs: song lyrics with section tags (such as verse, chorus, and bridge designations) and a structured caption describing the musical style and characteristics. By processing these inputs simultaneously in a single pass, the model eliminates the need for iterative generation steps, significantly accelerating the music creation process. The output maintains professional audio quality at 32 kHz sampling rate with 16-bit stereo resolution, meeting industry standards for music production.

The release includes three distinct serving paths, providing flexibility for different deployment scenarios and use cases. This modular approach allows users to integrate the model into various applications based on their specific technical requirements and infrastructure constraints.

• Democratization of Music Production: Open-weights distribution removes barriers for independent musicians, small studios, and developers lacking expensive licensing agreements
• Accelerated Creative Workflows: Single-pass generation from lyrics and captions streamlines music production pipelines compared to iterative alternatives
• Quality Standards: Professional-grade audio output quality suggests viable applications in commercial music production and content creation
• Integration Opportunities: Multiple serving paths enable flexible implementation across web applications, mobile platforms, and enterprise systems
• Research Accessibility: Availability to the academic community fosters innovation and development of improved music generation techniques

The release of MiniMax-Music3 addresses a critical gap in accessible, high-quality music generation technology. By combining open-weights distribution with impressive technical capabilities—particularly the ability to generate complete songs in single-pass processing—MiniMax enables a broader ecosystem of creators to leverage AI-powered music generation. This development signals the music production industry's continued evolution toward AI-assisted workflows, potentially reshaping how artists, content creators, and studios approach music composition and production in the coming years.