Model reference · open weights

stable-audio-open-small

stable-audio-open-small is an open-weight audio or speech model from Stability AI. stable-audio-open-small (FP32) weighs 2.5 GB; the smallest configuration that runs it is RTX 3060 12 GB.

• Stable Audio Open Small is a 497M parameter text-to-audio model developed by Stability AI.
• It generates variable-length stereo audio up to 11 seconds at 44.1kHz from English text prompts.
• The model uses a transformer-based diffusion architecture and is distributed under the Stability AI Community License.

Summary of the stabilityai/stable-audio-open-small model card, 2026-10-01

Hugging Face ↗GitHub ↗Paper ↗