Music app developers who want licensed models they can run or fine-tune themselves; three of the four are free open weights.
Kenning's take
What it does
Stable Audio 3.0 is a family of four audio models, from a small sound-effects model that runs on phones to a large music model for platforms. Small and Medium can make whole songs, with Medium and Large going past six minutes, and users own their outputs under Stability's Community License. Small SFX, Small and Medium are open weights on Hugging Face, while Large is available through the Stability AI API and enterprise self-hosting.
- Four models: Small SFX, Small, Medium and Large
- Full music composition on-device with 3.0 Small
- Tracks over six minutes with Medium and Large
- LoRA fine-tuning on your own audio library
- Audio inpainting and continuation
- Trained on fully licensed data
Who it suits. Developers, music apps and creators who want licensed music models they can run or fine-tune themselves.
Key facts
- Category
- AI models
- Made by
- Stability AI
- Based in
- United Kingdom
- Released
- 20 May 2026
- How to get it
- Small SFX, Small and Medium as free open weights on Hugging Face; Large through the Stability AI API and enterprise self-hosting
- Works on
- Hugging Face (open weights), API, On-device
- In the UK
- Yes
- Free plan
- Small SFX, Small and Medium are free open weights under the Stability AI Community License (organisations under M annual revenue)
Pricing
| Plan | Price | Billed |
|---|---|---|
| Enterprise License | contact sales Not stated | Not stated |
FAQ
When was Stable Audio 3.0 released?
Stability AI announced it on 20 May 2026.
Is Stable Audio 3.0 free?
Small SFX, Small and Medium are free to download as open weights. Organisations with over M in annual revenue need an Enterprise License.
Can Stable Audio 3.0 run on a phone?
Stability says 3.0 Small SFX and 3.0 Small are built to run on devices such as phones and consumer laptops.
Facts from the maker's own pages and the sources listed. Checked 2026-10-11.
Similar tools
ACE-Step 1.5
Open-source music generation model that writes full songs with lyrics and runs on ordinary consumer GPUs.
AlphaGenome Atlas
Google DeepMind database of AI-predicted effects for every possible single-letter change in human DNA.
Amazon Nova 2
Amazon's second family of Nova AI models, covering reasoning, speech conversation and combined text-and-image generation.
Cisco Antares
Cisco's small open-weight models that find where known security vulnerabilities sit in a codebase, running locally.