Stable Audio 2.5 from Stability AI delivers faster, more intricate AI‑generated music

Stability AI has released Stable Audio 2.5, an update to its audio generation model, aimed at enterprise-grade applications. The new version delivers higher quality, faster processing, and enhanced control. These improvements enable professionals to create dynamic musical compositions suited to custom branding requirements.

With a focus on the needs of creative teams, Stable Audio 2.5 allows users to generate customizable audio, including three-minute tracks produced in a matter of seconds. The model can also output multi-part compositions, supporting more complex audio arrangements for commercial use cases.

Building on its core workflows, Stable Audio 2.5 also introduces audio inpainting. This feature lets users upload their own audio, determine where their input begins, and have the model generate the remainder of the track using contextual cues. Both text-to-audio and audio-to-audio generation remain supported in this release.

Addressing enterprise concerns, the platform is commercially safe, trained exclusively on licensed data. Stable Audio 2.5 is accessible through Stable Audio's website, the Stability API, and integration partners such as fal, Replicate, and ComfyUI.