Suno v6 Launches With Warner, BMG and Believe as Development Partners

Two and a half years ago Warner Music Group was in court accusing Suno of training its models on Warner's catalog without permission. In November 2025 the two sides settled. On September 9, 2026, Warner announced it had gone from plaintiff to development partner: Suno launched Suno v6, a new generation of AI music models built with Warner, BMG and Believe — the first Suno models trained from scratch on licensed catalogs rather than on the unlicensed web data that triggered the industry's lawsuits. The release replaces the entire Suno model line at once, and it is the clearest signal yet that the streaming wars over AI music are being settled in boardrooms rather than courtrooms. Suno v6, Suno v6-wild and Suno v6-mini are the three tiers of the new generation.

This is not a normal point release, and it is worth being precise about what is actually new, because most coverage has focused on the celebrity of the partnerships and missed the structural change underneath. Suno v6 is not "v5.5 plus better vocals." It is a retrained model family whose training data, feature set, legal posture and even ownership terms are different from everything Suno shipped before it. If you have been generating music with Suno — or holding off on Suno for copyright reasons — the September 9 launch changes your calculation.

What shipped: three models, one generation

Suno retired every previous model the same day v6 launched. The model picker that used to offer v4.5, v5 and v5.5 now offers v6 and nothing else, which Suno says is the point: the platform is moving entirely onto the licensed generation rather than running old and new side by side.

• Suno v6 — the flagship, for Pro and Premier subscribers. Suno describes it as "powerful, versatile, refined," aimed at polished, genre-accurate songs from a text prompt. It is the model most people will hear first.

• Suno v6-wild — an experimental sibling, also Pro and Premier only. CEO Mikey Shulman says it is built for ideas rather than finished singles: "It may not make the perfect single song, but it will come up with some of the best ideas and some of the most interesting sounds." If v6 is the polished take, v6-wild is the one that surprises you, and Suno says it was developed "in much closer partnership with some of the best creatives in the world."

• Suno v6-mini — a free, faster, more efficient version available to every user, including free accounts. Suno's claim is characteristically large: that it "delivers better, faster results than any free model on any music creation platform." That is vendor marketing, unreproduced by any independent test at launch, but it signals where Suno is aiming the free tier.

Pricing did not move with the release. Suno's paid plans stay Pro at about $10 a month (less with annual billing) and Premier at about $30 a month (less with annual billing), per Suno's published US pricing, and both now include v6 and v6-wild. The free plan runs on v6-mini. Suno also showed a "Create Custom Model" beta in its walkthrough that builds a model from your own uploads for a fixed credit cost — a feature that was not in the launch blog post but is visible in Suno's own demo, so treat its exact pricing and availability as in flux.

The abilities that are actually new

v6's feature list is where the "not just a quality bump" claim gets concrete. The models add editing and sampling primitives that earlier Suno versions could not do, most of them aimed at the biggest practical complaint about AI music: you could generate a great first pass, but you could not fix the one bad bar without regenerating the whole song.

• Section-level rewrite. You can now point at a specific part of a track and change it in plain language — "make the chorus sung by a gospel choir" — without rebuilding the rest of the song.

• Single-lyric editing. Swap one word or line without regenerating the vocal performance around it.

• Mashups. Combine elements from multiple sources in one request: take the vocals from one track, the drums from another, and add new lyrics in a specified style.

• Sampling from audio. Drop in an existing recording and ask for a specific element — "sample the guitar riff at 0:45" — and build a beat or a song around it.

• Multimodal prompts. Text is no longer the only input. Images, video clips and voice memos can all drive generation, which makes v6 usable as a scoring tool for picture, not just a lyric-to-song machine.

• Advanced Mode. Beyond the old style sliders there are now controls for vocal gender, custom or auto duration, a Max Mode toggle for consistency across long songs, covers and voices, plus Weirdness, Style Influence and a new Variety slider, and a "My Taste" personalization setting. Generations are organized into workspaces rather than one flat library.

Suno claims the headline quality improvements — cleaner mixes, more controllable vocals, better prompt adherence — but at launch there are no independent listening scores for v6 anywhere. Everything in the previous paragraph is Suno-reported; the first real evaluation of v6 will be the community's, and it is coming within weeks. What can be verified today is the capability surface, which is genuinely larger than any prior Suno model.

The screen above is Suno's logged-out home page in the launch window: prompt entry plus the Studio/Resources nav that now sits on top of the v6-only generation stack. This is the product most v6 users will actually touch — a consumer studio, not a developer API.

Where the training data comes from now

The structural change is in the data. Suno's chief product officer, Jack Brody, said v6 "was trained entirely from scratch, from the ground up" on three sources: licensed catalogs supplied by the partners, user preference data, and Suno's own R&D. Warner chose which of its own catalogs to make available for training; BMG's music has been added; Believe signed its global licensing deal on September 8, the day before launch, bringing TuneCore artists and labels into the model. The preference data deserves emphasis because it is easy to misread: Suno says it is drawn from which of two generated versions users choose, not from uploaded audio.

Two absences matter. Universal and Sony are not in the training data — both are still suing Suno over the earlier, unlicensed models — and Brody said so directly. The German collecting society GEMA won a Munich ruling against Suno on July 31 that the company is still fighting. The v6 launch therefore does not make Suno's legal problems disappear; it makes the company's forward path licensed while its past remains in dispute.

Revenue is shared from day one. A portion of Suno's subscription revenue now flows to Warner, BMG and Believe, who decide among themselves how to pay artists and songwriters. Warner CEO Robert Kyncl said the company's artists "will share in the revenue generated by Suno's more than two million, and growing, paying subscribers" starting immediately. No percentages were disclosed.

Ownership terms also shifted. Suno says anything generated on v6 "is considered an original creation," and since September 3 its terms "assign" to Pro and Premier subscribers Suno's rights in outputs — while explicitly making no warranty that any copyright will vest in the output. Rights holders may still collect revenue from third-party distribution of what you generate. In plain terms: on the paid tiers you own what v6 makes for you, but "own" is a softer word in AI music than it is almost anywhere else, and the labels' slice comes out of the stream, not your subscription.

What comes next

Phase one is the models. Phase two, which Suno says arrives "in the coming months," is an opt-in artist program: individual artists choose to let fans remix or reimagine their catalog, and get paid when their music is used. That is the piece that would turn v6 from a licensed foundation into something closer to a two-sided marketplace, and it is also the piece that has historically been hardest to negotiate. More rightsholder deals are, in Brody's words, "absolutely our goal" — which is another way of saying the current three partners are the opening hand, not the final table.

For anyone building software rather than songs, the near-term takeaway is simpler: Suno v6 has no public developer API at launch. It is a subscription consumer product, served in Suno's apps, and the licensing deals run through Suno itself rather than through any reseller. The moment licensed music models do become callable over an API, teams that already build on a pass-through model router — one endpoint, provider list prices passed through with no markup — will be able to switch to them the day they appear instead of after a procurement cycle. For now, though, the fastest way to hear v6 is the same as it has always been: open Suno and prompt it.

The launch answers the question the music industry spent 2024 and 2025 fighting over — can AI music companies and rightsholders do business with each other — with a yes, but it leaves the harder question open. If v6's licensed models sound as good as Suno's lawyers and partners hope, the remaining lawsuits start to look like arguments about the past. If they sound a generation behind what the unlicensed models achieved, the industry just traded its legal problem for a quality problem. The first independent listening tests, and the first artist opt-in deals, will tell us which.