Stability AI Debuts Stable Audio 3.0 with Six-Minute Music G

Stability AI has released Stable Audio 3.0, a significant update to its generative music platform that now allows users to create full-length tracks up to six minutes long. This version doubles the previous three-minute limit found in version 2.0, enabling the generation of complete songs with distinct structural elements like intros, verses, and choruses.

The new model is built on an improved latent diffusion architecture designed to handle complex musical arrangements. Beyond the extended duration, Stable Audio 3.0 introduces native awareness of song structure, which helps the AI maintain musical consistency across longer runtimes. This development is particularly useful for creators who need background music or thematic scores that do not require manual looping or extensive editing.

For users with high-performance hardware, the update brings support for local inference. Stability AI confirmed that the model is optimized for NPU-equipped PCs, allowing for faster processing directly on a user's device rather than relying solely on cloud servers. This shift toward on-device AI reflects a broader trend in the industry to reduce latency and improve privacy for content creators.

Commercial availability is also changing with this release. Stability AI is providing open weights for Stable Audio 3.0 to members of its Pro commercial tier. This allows professional users to integrate the technology into their own workflows or host the model on their private infrastructure. As of May 21, 2026, the tool is available for testing through the official Stable Audio website, where users can experiment with the new six-minute generation capabilities.

While we strive for accuracy, bytevyte can make mistakes. Users are advised to verify all information independently. We accept no liability for errors or omissions.

Sources

Stable Audio 3.0: Full-Length Music Generation

AI-generated image.

✔Human Verified

Sources

Related Articles