Suno has launched v6, its first artificial intelligence music model trained using data licensed from major record labels. The software company confirmed that the new system draws from partnerships with Warner Music Group, BMG, and Believe, alongside user-generated content. Jack Brody, a Suno executive, stated the model is trained from the ground up on a distinct dataset different from previous versions. This shift marks a departure from earlier iterations which relied heavily on publicly scraped internet data. The release introduces three specific variants: v6, v6-wild, and v6-mini. The mini version remains free for all users and prioritises speed.
This development matters because it represents a structural change in how the company sources training material. By integrating formal licensing deals, Suno attempts to address copyright concerns that have plagued the industry. The inclusion of licensed content suggests a move away from the legal grey areas associated with unauthorised scraping. However, the presence of user data in the mix means the training set is not entirely composed of professional recordings. This hybrid approach could influence future legal challenges regarding the ownership of generated music.
* v6-mini is free and optimised for quick generation
* Training data combines licensed tracks with user uploads
* Warner Music Group, BMG, and Believe are key partners
