Stability AI's new audio model can now create 6-minute songs

Making music just got easier! Stability AI just launched Stability Audio 3.0, and the top version can create professional music over 6 minutes long—more than double what the previous model could do.

There are four models in total. The small ones run on your phone and make audio up to 2 minutes. The big one, with 2.7 billion parameters, can create a full 6-minute, 20-second song while keeping the melody and structure perfect. Great for musicians and content creators.

Stability AI open-sourced the small and medium models. But the best one is only available through API or paid service. Companies making over $1 million a year need a special license. All models are trained on legal music data, thanks to deals with Warner Music and Universal Music Group.

If you make short videos, podcasts, or game music, try the free medium model first—it's cheap and sounds great.