Stable Audio
Stability AIStable Audio is Stability AI's family of diffusion models that generate music and sound effects from text. The Stable Audio 3 generation ships Small Music, Small SFX and Medium models that produce clips of up to two and six minutes and support editing and continuation, alongside the earlier Stable Audio Open models. The largest Stable Audio 3 model is not publicly downloadable; Stability offers it through its API and enterprise self-hosting.
The stable-audio-tools training and inference library is not part of this entry; it is a separate library in the registry.
Openness
3 high confidence- weights
- open(the Stable Audio 3 Small and Medium checkpoints and Stable Audio Open download from Hugging Face, some behind an automatically granted click-through
- data
- described(trained on licensed and Creative Commons data, which is described and not released)
- code
- open(the training and inference pipeline is published in Stability-AI/stable-audio-3 under MIT)
- license
- Stability-AI-Community-License(every distributed checkpoint: free below USD 1M annual revenue, after which an enterprise license is required)
Every downloadable Stable Audio model is free to use, commercially included, for organizations under USD 1M in annual revenue, with a paid license above that. The training pipeline is published and the licensed training data is described but not released. Stability-AI-Community-License allows commercial use only within a bound, so the license tier is use_bounded.
- https://cdn.jsdelivr.net/gh/Stability-AI/stable-audio-3@main/LICENSE recorded 2026-09-26
stable-audio-3 code LICENSE: "MIT License Copyright (c) 2026 Stability AI".
- https://huggingface.co/api/models?author=stabilityai&search=stable-audio&sort=downloads&direction=-1&limit=100 recorded 2026-09-26
Hugging Face listing of the stable-audio repositories: the Stable Audio 3 and Stable Audio Open checkpoints are public.
- https://huggingface.co/stabilityai/stable-audio-3-medium-base/raw/main/LICENSE.md recorded 2026-09-26
License body: "STABILITY AI COMMUNITY LICENSE AGREEMENT Last Updated: July 5, 2024", preserving "free access to the Models for people or organizations generating annual revenue of less than US $1,000,000".
- https://huggingface.co/stabilityai/stable-audio-3-medium-base/raw/main/README.md recorded 2026-09-26
Card: "Stable Audio 3 models are trained on licensed and Creative Commons data" and "We release the weights of small and medium ... together with their training and inference pipeline."
- https://stability.ai/news-updates/meet-stable-audio-3-the-model-family-built-for-artistic-experimentation-with-open-weight-models recorded 2026-09-26
"3.0 Small SFX, 3.0 Small, and 3.0 Medium are all open-weights." and "Stable Audio 3.0 Large is available via the Stability AI API and self-hosting for enterprise deployments."
Adoption
3 high confidenceAdoption is measured as Hugging Face downloads across the nine generation checkpoints, led by Stable Audio 3 Medium. The separate autoencoder repositories are left out.
- https://huggingface.co/api/models?author=stabilityai&search=stable-audio&sort=downloads&direction=-1&limit=100 recorded 2026-09-26
Listing sorted by downloads: stable-audio-3-medium 81781, stable-audio-3-small-sfx 28976, stable-audio-open-1.0 20764 in the trailing 30 days.
Capability
3 medium confidenceStable Audio 3 Large places in the upper half of the instrumental music arena, and the downloadable Medium model in its lower half. Suno and Lyria both place above the family's best model.
- https://artificialanalysis.ai/music/leaderboard/instrumental recorded 2026-09-26
Arena leaderboard "AA-Music-Instrumental v1.1": "Stable Audio 3 Medium" (rank 13, Elo 1000.3), "Stable Audio 3 Large" (rank 8), "MusicGen" (rank 19).
Verified 2026-09-26