Tier 2 Restricted open-weight
What the publisher claims: open-weights, open models
The one-liner: Open audio weights under the same community license that governs Stability's image releases.
The six questions
Weights for the Open and 3.0 families are downloadable from the publisher's repositories, gated only by agreement acceptance.
Commercial and non-commercial use are permitted under the Community License, but organizations above $1M in annual revenue must move to enterprise terms.
Self-hostable from the published weights and inference code.
Papers and release notes describe the models; the training audio corpus and recipe are not published.
Limited.
Stability-AI/stable-audio-3 and stable-audio-tools accept the odd outside PR (only two non-bot external merges in the model repo), and Stability revises its Community License unilaterally, so the audio roadmap is a one-way corporate release.
License
Stability AI Community License — license source ↗
- Organizations with more than $1M in annual revenue must contact Stability AI for enterprise licensing
- Published releases carry a license tag of 'other' rather than a named OSI licence, and the repositories are gated by agreement
- Stability AI has repeatedly revised the audio release terms
What is actually released
- weights (Open 1.0, Open Small, 3.0 Small, 3.0 Medium, 3.0 Small SFX)
- papers
- model cards
- hosted API endpoints
Why it matters
Extends the Stability AI pattern from images into audio, showing that one community-license design can cover a whole product line across modalities.
Sources
- https://huggingface.co/stabilityai/stable-audio-open-1.0
- https://huggingface.co/stabilityai/stable-audio-3-medium
- https://releasebot.io/updates/stability-ai
Derived from the openness dataset behind But Is It Open? — snapshot 2026-11-07 ↗. Scores are qualitative readings of the cited evidence, not verdicts; tier measures openness, not intelligence.