Stability AI’s newest audio model can generate songs that run longer than many radio singles.
The company has released Stable Audio 3.0, a family of AI audio models that can create music and sound effects, with larger versions capable of generating full compositions up to 6 minutes and 20 seconds long.
Stable Audio 3.0 goes longer
According to TechCrunch, Stability AI is releasing four models under the Stable Audio 3.0 name: Small SFX, Small, Medium, and Large. The two Small models have 459 million parameters and are built for on-device sound and music generation of up to two minutes.
The Medium and Large models stretch the output much further. Both can create full compositions of six minutes and 20 seconds while maintaining musical structure and melodic tone. The Medium model has 1.4 billion parameters, while the Large model has 2.7 billion.
Stability AI’s own Stable Audio 3.0 announcement says the Small SFX, Small, and Medium models are open weights. The Large model is available through the Stability AI API and self-hosting for enterprise deployments.
The release also updates Stability AI’s audio lineup. Stable Audio Open, released in 2024, could generate up to 47 seconds of samples and sound effects. Stable Audio 3.0 moves the company into longer-form music generation while keeping smaller models available for experimentation and local use.
Stability AI said all Stable Audio 3.0 models were trained on fully licensed data. Under the Stability AI Community License, users own their outputs and can distribute and commercialize them. Organizations with more than $1 million in annual revenue need an enterprise license, which Stability AI says includes commercial coverage and legal indemnification.
AI music tools are arriving during an active copyright fight, with creators and lawmakers pushing for more transparency around training data.
Open models, paid access, and music-industry ties
Stable Audio 3.0 is not just one model with one audience. Stability AI is splitting the family across creators, developers, and larger companies that need API access or self-hosting.
The company said the Small SFX model is designed for sound effects on devices such as mobile phones and consumer-grade laptops. The Small model is built for full music composition on-device. The Medium model offers longer tracks and stronger musicality, while the Large model is designed for music platforms and creative applications that require low-latency generation at high volume.
The new architecture also supports editing tools. Stability AI said Stable Audio 3.0 can handle audio inpainting, continuation, and variable-length generation with per-second control. That means users can modify part of a track, extend a composition, or generate a shorter section without having to start over.
Stability AI is also trying to build credibility with the music industry. TechCrunch reported that the company signed deals last year with Warner Music Group and Universal Music Group to develop models and music-creation tools. The company is also building a product suite for professional musicians, with Ethan Kaplan, former chief digital officer at Universal Audio and Fender, joining to lead that effort.
Stable Audio 3.0 arrives as Google is pushing deeper into AI music with Gemini and Lyria tools, and as audio becomes a larger part of the AI product race, from assistants to rumored audio-first AI devices.
For creators, the appeal is longer output and more control. For companies, the draw is a model family that combines open-weight experimentation with paid enterprise access.
The model family gives both groups a path toward faster audio creation without losing sight of licensing and ownership.
Also read: OpenAI’s latest voice AI update shows how quickly companies are turning voice tools into systems that can listen, respond, translate, and act.