Stable Audio 2.5
Audio modelPrevious generationCard summarizedThe first audio generation model designed specifically for enterprise-grade sound production
Source: Stable Audio 2.5 launch post and Stable Audio 2.5 API reference (no model card published)
Where to call it
- Stability AI Platform API
stable-audio-2.5
Good for
- Enterprise sound production at scale
- Music production, cinematic sound design and remixing
- Multi-part compositions with intro, development and outro
Overview
Stable Audio 2.5
The first audio generation model designed specifically for enterprise-grade sound production
- Max length
- 190 s
- Languages
- English prompts only
- Modes
- text-to-audio (music and sound effects), audio-to-audio
- Released
- 10 Sep 2025
Stable Audio 2.5 is Stability AI's hosted music and sound model, launched on 10 Sep 2025 and aimed at enterprise sound production. It generates 44.1 kHz stereo tracks up to about three minutes with multi-part structure (intro, development, outro), transforms uploaded audio and inpaints sections. Stability says it generates a three-minute track in under two seconds on a GPU. It was post-trained with Stability's ARC method and trained on licensed data. Stable Audio 3.0 has since replaced it as Stability's most advanced audio model, at the same price. launch post; API reference
stable-audio-2.5- Typical price
- $0.08 / song
- Max length
- 190 s
- Languages
- English prompts only
- Weights
- Closed
- Released
- 10 Sep 2025
- text-to-audio (music and sound effects)
- audio-to-audio
- audio inpainting