Stability AI Unveils Stable Audio 2.5, Built for Enterprise-Grade Sound Production

Stability AI has launched Stable Audio 2.5, the first audio generation model purpose-built for enterprise sound production at scale. The release marks a major step in addressing the growing demand for customised, high-quality audio across brand touchpoints — from advertising to in-store experiences.

Unlike prior consumer-focused audio models, Stable Audio 2.5 is designed for professional creative teams, delivering three-minute tracks in under two seconds on GPU and supporting advanced workflows such as text-to-audio, audio-to-audio, and audio inpainting.

"With the launch of Stable Audio 2.5, we’re also partnering with a leading sound branding agency to co-develop enterprise solutions for innovative brands that want to create iconic sound identities and experiences," the company said in a press release.

Built using the company’s proprietary ARC (Adversarial Relativistic-Contrastive) method, the model produces dynamic, multi-part musical compositions while adhering closely to mood and genre-specific prompts. Stability AI emphasizes that the system is commercially safe, trained on a fully licensed dataset.

“Custom audio can make a brand eight times more memorable, but only 6% of creative uses a sound identity,” the company noted, citing Ipsos research. Stable Audio 2.5 is designed to help enterprises close that gap by creating distinct, brand-led soundscapes adaptable across platforms.

To accelerate adoption, Stability AI is partnering with amp, a Landor Group company under WPP, to co-develop enterprise sound solutions. Through WPP Open, Stable Audio 2.5 will be offered to the agency’s global client base, blending cutting-edge generative AI with creative expertise.

Stable Audio 2.5 is now available at StableAudio.com, via the Stability AI API, partner platforms such as Replicate, fal, and ComfyUI, and with enterprise on-premises licenses.