Stable Audio 3.0 🧠 Model Open source

What it is

Stable Audio is Stability AI's music and sound-effects family. Stable Audio 3.0, announced on 20 May 2026, is a model family trained on fully licensed data. Large is an enterprise model. Medium, for full songs, and Small / Small SFX, for mobile and sound effects, are released as open weights. Stability's investors now include Warner Music Group, Universal Music Group and Sony Music.

Key features

• Tracks up to six minutes long
• Music across genres, plus dedicated sound-effect models
• Edit a segment, rework part of a song, or extend a composition
• Medium generates in about 2 seconds on an H200, and a few seconds on a MacBook Pro M4
• Trained on fully licensed data, with indemnity under the Enterprise licence

Where you can use it

The Stable Audio web app (stableaudio.com), the Stability AI Platform API and self-hosted enterprise licences. The open Small and Medium weights are on Hugging Face and can run locally.

Pricing and rights

Open weights come under the Stability AI Community License. Stability sends commercial users to its licence page, so check your revenue tier there. Medium also bundles a T5Gemma text encoder under Gemma's terms. We could not load the app's subscription prices. Enterprise customers get legal indemnification.

Who it is for

Brands, game studios and video editors who need safe-to-license background music and SFX, and developers who want an open audio model to run on their own hardware.

Verdict

Stable Audio 3.0 bets on licensed training data, which is its main selling point for commercial buyers. It does not make songs with vocals like Suno or Lyria, and app pricing plus the community-licence terms need checking before scaled use.