Suno has begun rolling out v6, its first AI music model developed with support from record-industry partners. The company says it trained the system from scratch on a new dataset that includes licensed material from Warner Music Group, BMG and Believe, as well as user data. Suno has not clarified whether the training set excludes every source disputed in connection with its earlier models.
The release comes in three versions. Standard v6 is the main paid model, v6-mini is a faster and less resource-intensive option available free to everyone, and v6-wild is intended to produce less predictable results. A hands-on test by The Verge found stronger recognition of genre cues, but also more vocal artifacts than v5 in some outputs and little success when explicitly requesting off-key or rhythmically imperfect music.
Users can now revise part of a song through a chat interface, changing an instrument line or lyric without regenerating the whole track. The model can also combine elements from a user’s Suno library and accept images, video or audio as prompts instead of relying only on text.
Suno says it will eventually retire its older models. The launch advances licensed AI music, but questions remain about user-data rights, the full composition of the training set and how well the model follows unusual musical instructions.