Stability AI has introduced Stable Audio Open Small, a compact audio-generating AI model built to run directly on mobile devices. In partnership with chipmaker Arm, this new release marks a bold step toward offline, real-time music generation. Unlike existing tools like Suno and Udio, which rely heavily on cloud servers, Stability’s new model can generate stereo audio on smartphones in under eight seconds.
At just 341 million parameters, Stable Audio Open Small is engineered for speed and efficiency. It runs natively on Arm CPUs—the processors found in most phones and tablets—enabling developers and creators to produce quick sound effects and music clips without an internet connection. Use cases include fast riff generation, drum loops, or ambient sound design for mobile games and creative apps.
One of the standout features of Stable Audio Open Small is its transparent training data. Stability claims the model was trained entirely on open-source and royalty-free tracks from Free Music Archive and Freesound. That’s a stark contrast to rival audio models like Suno and Udio, which have faced criticism over possible use of copyrighted material. This clear-cut approach helps developers sidestep legal uncertainty and builds trust in the model’s outputs.
However, there are trade-offs. Stability notes the model’s audio quality varies across genres due to the Western-centric dataset. It also struggles to produce realistic vocals or full-length, high-quality songs. Another limitation: it only accepts English prompts.
Free to Experiment, But Commercial Use Has Limits
The launch comes amid a turbulent time for Stability AI. The company is still recovering from a rocky year marked by executive shakeups, missed partnerships, and financial strain. Founder Emad Mostaque exited after internal turmoil, and recent investors like Eric Schmidt and Sean Parker are backing efforts to refocus the business.
Today, Stability has a new CEO, filmmaker James Cameron on its board, and a wave of new releases across image and audio generation. Still, not all is open source: Stable Audio Open Small is free for researchers, hobbyists, and small businesses making under $1 million annually. But for developers or companies earning more, an enterprise license is required.
By blending efficient audio generation with royalty-free training and mobile support, Stable Audio Open Small could be the entry point for next-gen music tools built entirely offline.
