Stability AI has launched Stability Audio 3.0, a family of audio models that significantly extend the length and quality of AI-generated music, with the top models producing compositions over six minutes long.
- Top models create music over six minutes long
- Open weights for smaller models enhance developer access
- Commercial licenses required for large-scale use
What happened
Stability AI, known for its work on Stable Diffusion, has announced Stability Audio 3.0, a new suite of audio generation models that can produce music tracks lasting up to six minutes and 20 seconds. The release includes four versions: two small models optimized for on-device use and shorter tracks of up to two minutes, and medium and large models capable of longer compositions maintaining musical coherence.
This update represents a substantial leap from the company's 2024 Stable Audio 2.0, which could generate tracks up to just under three minutes long. The company is making all but the largest model open-source, supporting broader experimentation and development, while reserving enterprise licensing and paid API access for the most powerful large model.
Why it matters
The advances in Stability Audio 3.0 push the boundaries of AI-generated music by offering more professional and structurally sound compositions at greater lengths. This holds potential for creators, developers, and musicians seeking new tools for music production, especially with open access to several models enabling innovation without high barriers to entry.
In an industry marked by concerns over data licensing and copyright, Stability AI’s commitment to using fully licensed data and partnerships with major music labels like Warner Music Group and Universal Music Group is significant. It demonstrates a pathway for AI music companies to operate responsibly while expanding capabilities, amid ongoing legal and commercial debates around AI training data.
What to watch next
Stability AI plans to expand its professional music offerings, recently hiring industry experts to lead development efforts, which could signal new products tailored for musicians and commercial music production. Observers will be watching how these tools integrate with existing workflows and how the company navigates licensing and enterprise customer demands.
The broader AI music sector continues to evolve rapidly, with competitors like Google, ElevenLabs, and Suno growing their own technologies and business models. The balancing act between innovation, legal compliance, and commercial viability will likely shape which platforms gain traction and partnerships moving forward.