Stability AI launches Stable Audio 3.0, capable of generating six-minute songs

Stability AI, the company best known for its Stable Diffusion image generator, has released a new family of audio models called Stability Audio 3.0. The company claims the top-tier model can generate professional-grade music tracks lasting more than six minutes, a significant leap in capability for AI-generated audio.

Four new models, varying in size and capability

The release includes four distinct models under the Stable Audio 3.0 banner. Two small models, named small SFX and small, each contain 459 million parameters and are designed for on-device sound effects and music generation of up to two minutes. The medium model, with 1.4 billion parameters, and the large model, with 2.7 billion parameters, can both create full compositions lasting six minutes and 20 seconds. Stability AI says these longer tracks can maintain musical structure and melodic tone throughout, addressing a common limitation of earlier AI music generators.

This represents more than double the generation length of Stable Audio 2.0, which was released in 2024. The previous open-source version, Stable Audio Open, could only generate up to 47 seconds of music.

Open weights and licensing strategy

Stability AI is making the small SFX, small, and medium models available with open weights, meaning developers and researchers can download, use, and modify them freely. The large model, however, is only accessible through the company’s API and self-hosted paid services. Additionally, companies with annual revenue exceeding $1 million will need to obtain an enterprise license to use the larger model commercially.

Licensed data and music industry partnerships

The AI startup has emphasized that its latest audio models are trained on fully licensed data. This is a critical distinction in the current AI music landscape, where companies like Suno and Udio are facing ongoing copyright lawsuits from major record labels over the use of unlicensed music for training data. Stability AI previously signed deals with Warner Music Group and Universal Music Group in 2024 to develop models and music creation tools, giving it a stronger legal foundation than some competitors.

In a move that signals deeper commitment to the professional music market, Stability AI has hired Ethan Kaplan, former chief digital officer at Universal Audio and Fender, to lead its professional music offering. The company is developing a new suite of products for professional musicians, though it has not yet provided specific details about features or release dates. This hiring trend is visible across the industry: Suno recently brought on former Merlin CEO Jeremy Sirota as chief commercial officer, and ElevenLabs hired Derek Cournoyer from indie music publisher Kobalt as a strategy lead for its music business.

Why this matters for creators and the industry

The ability to generate coherent, structurally sound music tracks of over six minutes opens new possibilities for content creators, game developers, and independent musicians who may lack the budget for custom compositions. However, the legal and ethical questions surrounding AI-generated music remain unresolved. The outcome of the Suno and Udio lawsuits, combined with the licensing agreements Stability AI has secured, will likely shape how the music industry approaches AI tools going forward. For now, Stability AI’s move positions it as a more legally cautious player in a rapidly evolving field.

Conclusion

Stability Audio 3.0 represents a notable technical advancement in AI music generation, doubling the output length of its predecessor while maintaining musical quality. The company’s strategy of offering open-weight models alongside licensed training data and high-profile music industry hires suggests a deliberate effort to build credibility with both developers and the traditional music business. As the legal landscape around AI-generated music continues to develop, Stability AI’s approach may serve as a template for balancing innovation with compliance.

FAQs

Q1: How long can Stable Audio 3.0 generate music?
The medium and large models can generate full compositions of up to 6 minutes and 20 seconds. The small models are limited to two minutes.

Q2: Is Stable Audio 3.0 free to use?
The small SFX, small, and medium models are available with open weights for anyone to use and modify. The large model is only available through paid API access or self-hosting, and companies with over $1 million in revenue need an enterprise license.

Q3: How is Stability Audio 3.0 different from other AI music generators like Suno or Udio?
Stability AI has focused on using fully licensed training data, having signed deals with Warner Music Group and Universal Music Group. This contrasts with Suno and Udio, which are currently facing lawsuits over alleged use of unlicensed music for training.