AI-powered music platform Suno has launched v6, its first model family developed with the music industry, following agreements with Warner Music Group, BMG and Believe. According to Chief Product Officer Jack Brody, the model was trained on a new dataset created using content licensed from its partners and user data, without using previous data; the dataset was not disclosed in full.
The family includes v6 for controlled results, v6-Wild for experimental generations, and the faster v6-mini, which is available to all users for free. The first two models are available to Pro and Premier subscribers. The Verge’s tests found that the mini could produce simpler outputs and more audio distortion, while v6 understood hyperpop and krautrock prompts better but struggled to generate natural performance imperfections.
Users can change sections of tracks through chat, combine outputs in their library, and generate music from images, video or audio files. Suno plans to retire its older models. The company is also working on licensed AI products through which artists who opt in will earn revenue.
Why it matters
This development shows that in AI music generation, not only audio quality but also the data and rights holders used to develop the models have become fundamental elements of the product. The use of licensed content makes relationships with music companies more visible, while the failure to disclose the full dataset leaves an open question regarding transparency and rights holders’ oversight. Separating different models into paid and free tiers establishes a direct link between the level of generation users can access and the quality of their experience; the distortions observed in testing also show that this difference is not merely a marketing choice. The plan for artists to earn revenue through optional participation, meanwhile, keeps the question of how revenue arising from the widespread adoption of generation tools will be shared on the agenda.