Stability AI, the company behind Stable Diffusion, is releasing a new family of audio models, called Stability Audio 3.0. The top model can generate professional-grade music of more than six minutes long, the company claimed.

The company is releasing four new models under the Stable Audio 3.0 name: small SFX (459M parameters), small (459M parameters), medium (1.4B parameters), and large (2.7B parameters). The duo of small models is suitable for on-device sound and music generation of up to two minutes.

Both medium and large models can create full compositions of 6 minutes, 20 seconds long that can maintain musical structure and melodic tone. This is more than double the length of what Stable Audio 2.0, released in 2024, was capable of generating.

Stability AI is making small SFX, small, and medium models available with open weights for anyone to use and modify. In 2024, the company released Stable Audio Open, which allowed for music generation of up to 47 seconds. The new family of models is a big step up from the previous open versions.

The large model is available only through the API and self-hosting paid services. Plus, companies with more than $1 million in revenue would need to get an enterprise license.

Many companies, including Google and ElevenLabs, are releasing models and tooling around music generation. However, as Suno’s and Udio’s ongoing court battles have proved, licensing of data and partnerships with music labels could become a key part of the long-term survival of these services.

Last year, Stability AI inked deals with Warner Music Group and Universal Music Group to develop models and music-creation tools. The company said that its latest set of audio models is built on fully licensed data.

The AI startup is developing a new suite of products for professional musicians but didn’t give more details on its features. Ethan Kaplan, former chief digital officer at Universal Audio and Fender, is joining the company to lead Stability’s professional music offering.