Stability AI releases Stable Audio 3.0

stability ai introduced stable audio 3.0, with the medium and large models capable of maintaining musical structure and melodic tone for over six minutes. the open-weight small, small sfx, and medium models allow local community use, while the large model requires a paid license for companies making over $1 million in revenue. the models are trained exclusively on licensed data from partnerships with warner music group and universal music group to avoid ongoing copyright litigation. ethan kaplan, former chief digital officer at universal audio, joined stability ai to lead professional music products.