Suno shipped v6 on September 9. Three models, trained from scratch, and a training set that Suno describes as a combination of data licensed from these partners, user data, and the accumulation of the team's learnings (Source).
The licensing partners are Warner Music Group, BMG, and Believe, the distributor whose catalog includes TuneCore artists. Warner settled a copyright lawsuit with Suno in November 2025. BMG reached its deal more recently, per reporting from Music Business Worldwide and TechCrunch. The two holdouts from the major labels, Universal Music Group and Sony Music Entertainment, are not part of the arrangement. Their lawsuits, filed under RIAA coordination in Boston in June 2024, remain active.
That context is the frame through which v6 needs to be understood. The story here is an industry shift, not an isolated first. Udio had already moved toward the licensed model before v6 arrived, signing deals with Universal Music Group, Warner Music Group, Merlin, and Kobalt after settling UMG's lawsuit in October 2025. Rights holders are now being paid on both sides of the category.
What Ships With v6
Three models. The flagship is simply called v6, available to Pro and Premier subscribers. A second model, v6-wild, carries the same access tier and is described by Music Business Worldwide as "less predictable and more varied." A third, v6-mini, is the free-tier model and the only one available without a paid plan.
The Suno pricing page shows three plans. Free is $0, with 50 credits per day and no commercial rights. Pro runs $10 per month on a monthly billing cycle or $8 per month billed annually, comes with 2,500 credits monthly and 20 song downloads, and includes commercial use rights (Source). Premier is $30 per month ($24 per month billed annually) and includes 10,000 credits, 60 downloads, and Suno Studio with MIDI, effects, and automation (Source). Pro and Premier subscribers get v6 and v6-wild access. Free users get v6-mini.
On the feature side, the main addition is natural language editing. Suno's Music Ally coverage cites the example: "Change the chorus so it's sung by a gospel choir." Text, images, and video are also supported as creative references. Suno said last month that it plans to add a watermark to songs generated using its platform (Source). A remix feature is planned but not live at launch, with an artist opt-in model described in press materials.
Suno's chief product officer, Jack Brody, stated in press coverage that revenue sharing with partners begins at launch.
Why the Training Data Question Matters
Every prior version of Suno was trained on recordings that rights holders say they never authorized for this purpose. That is the core allegation in the litigation that has accumulated around the company. TechCrunch reported that the company "admitted to training its models using YouTube videos" a day before the v6 announcement (Source).
The result has been a lawsuit environment that has complicated the entire AI music generation category, not just Suno. GEMA, the German music rights organization, won a copyright ruling against Suno at Munich Regional Court on July 31, per Music Business Worldwide coverage. Class action suits from individual artists, including Jason Isbell, are pending. SOCAN and Round Hill Music have separate claims. Universal and Sony are still in the RIAA-coordinated case in Boston.
Suno has raised over $819 million in total funding, per TechCrunch. That is a meaningful runway for litigation. But the legal exposure has been a legitimate concern for any business or creator thinking about commercial use of AI-generated music. If a court rules that the underlying training data was improperly used, what happens to the music you made with it? That question has no clean answer under previous model versions.
v6 does not resolve the active suits, which cover previous models. What it does is establish a different starting point for future output: v6 was trained partly on licensed source material, alongside user data. Whether that matters legally or commercially depends on how the remaining cases resolve and how seriously users weigh the distinction.
For the portion of the AI audio and music space that has held off on commercial deployment because of training data uncertainty, v6 removes at least one layer of that concern. Whether Warner, BMG, and Believe constitute a comprehensive enough sample of music history to produce a model with real range is a separate question. Sony and Universal catalogs are not in the training set.
Read the rest of this article on TechSifted →
Originally published at techsifted.com.
Top comments (0)