{"slug": "suno-rolls-out-v6-song-making-model-will-pay-royalties", "title": "Suno rolls out v6 song making model, will pay royalties", "summary": "Suno launched its v6 AI music model family on [date not specified], trained from scratch on a dataset that includes licensed music from Warner Music Group, BMG, and Believe, while retiring all previous models. Chief product officer Jack Brody called v6 'our most powerful music creation tools to date,' with three tiers: v6 for paying subscribers, v6-mini for free users, and v6-wild for exploration. The launch comes amid ongoing lawsuits from UMG, Sony Music, and others over prior training data.", "body_md": "Today is a big day for Suno: it is launching its latest AI-music models. Suno v6 is the first generation developed with some rightsholders – [Warner Music Group](https://musically.com/2025/11/25/ai-music-firm-suno-strikes-first-licensing-deal-with-warner-music-group/),  [BMG](https://musically.com/2026/08/12/suno-bmg-ai-music-licensing-deal/) and yesterday’s addition [Believe](https://suno.com/blog/believe-partnership) – as licensed partners.\n\nSuno is still facing lawsuits from major labels UMG and Sony Music, as well as a growing list of other litigants that includes [musicians](https://musically.com/2026/09/02/suno-sued-by-musicians-including-david-lowery-and-jason-isbell/), Canadian collecting society [Socan](https://musically.com/2026/09/03/canadas-socan-files-latest-music-industry-lawsuit-against-suno/) and rights owner [Round Hill Music](https://musically.com/2026/08/18/ai-music-roundup-round-hill-lawsuits-alibaba-submithub-and-more/).\n\nThat makes v6 an important moment for Suno though: the first proper look we’re getting at how its service will evolve with proper licensing deals and industry partners.\n\nHere’s the basic information. v6 is a family of three models, although one of them is just called ‘v6’. That’s the best-quality version of Suno that will only be available for paying subscribers, while ‘v6-mini’ is the version for free users, which is also billed as being faster and more efficient.\n\nAnd then there’s ‘v6-wild’ which Suno describes as being “built for exploration… less predictable and more varied, producing unexpected, textured and ambitious results”.\n\nThat’s something we’ll explore more in this breakdown of what you *really* need to know about Suno v6, based on a pre-launch interview with chief product officer Jack Brody.\n\n“These are our most powerful music creation tools to date. They are faster, they’re more expressive, they’re higher quality, and they bring a bunch of new features that ultimately offer greater control over the music people are creating,” is how he described v6.\n\n## Suno v6 is ‘trained from scratch’\n\nHere’s the first key point: “v6 is trained from the ground up, so it’s trained from scratch on a new set of data that is not the same data that previous models were trained on,” said Brody.\n\n“This includes licensed music from Warner. It includes user data, and then it includes all of these learnings and innovations that the team has has made.”\n\n(Since the interview, Suno tells Music Ally that BMG music has begun being added into v6, and that music from participating Believe and TuneCore artists will also be included in the new models.)\n\nSo: brand new models, a brand new dataset. And to emphasise the idea of a clean break from its past, the v6 launch also sees Suno retiring all of its previous models. That’s not new news today: [it announced the retirement plans in August](https://suno.com/blog/suno-updates-tos).\n\n“I think of this moment as really marking a new chapter for Suno,” was how Brody put it.\n\nRetiring previous models and debuting a new one trained on new data is not a get-out-of-lawsuits-free card for Suno. The litigation based on those previous models continues.\n\nIt feels like more of an attempt to create a clean legal boundary between ‘old’ and ‘new’ Suno, focusing the legal battles on legacy models that don’t exist on the platform – even if the “user data” used in v6’s training dataset does come from those days.\n\nDoes having properly-licensed music from labels *improve* Suno’s model? That’s a thorny question. Remember, those lawsuits hinge upon accusations that Suno’s pre-v6 models were trained on labels’ music without permission. But Brody also suggested that the deals are not the main factor in Suno getting better.\n\n“The wrong way to think about these partnerships is that they are about the data, or even specifically improving the model,” he said.\n\n“The the vast majority of what makes a model great at this point is the technical breakthroughs of the team, the taste of the team, the accumulation of of learnings over many years of building these things, and the the feedback we get from artists and users.”\n\n## Royalty payouts: yes. Remixes: soon…\n\nv6 does mark a new chapter in Suno’s relationship with music rightsholders though: from today, it will be paying them.\n\n“When we roll out these models in v6, that’s actually when we are officially rolling out this partnership with WMG, and now with BMG as well,” said Brody. “From the jump, as we start to roll out this partnership, we will start sharing a portion of our revenue with our partners.”\n\n“The big story here is that from the beginning, we’ve always believed that these creative tools that empower more people to make music will grow the overall pie for the music industry. And we already see this,” added Brody.\n\n“When people start paying to use Suno, they don’t stop paying for Spotify: they’re just paying more for music. And this [v6] marks the beginning of the chapter where a portion of that revenue – and incremental new streams that we will generate with new products that we build as a result of this partnership – will start playing out.”\n\nSo, v6 represents the start of Suno paying out royalties to the music industry. However, what it doesn’t represent *yet* is the other part of the company’s plans for its licensing deals.\n\n“Ultimately, what this product partnership really is about is building with the industry and enabling new product experiences. So although the partnership launches when v6 does, it really starts to come into effect as we start to roll out some of these new product experiences,” said Brody.\n\n“The first version of that is going to be an experience where artists can opt in, and then their likeness and catalogue can be used for remix experiences, which has always been one of the on-ramps to music creation.”\n\n“People take familiar music and they cut it up and remix it, but on Suno it’s always been about original creation, so that [remixing] has never been a capability we’ve offered. We’re excited to bring that into the platform, with consent from these artists, and in turn with the ability to pay them through their labels and rightsholders.”\n\nIn short, the remix tools are still being developed, but they’re not going live with v6 today. We’ll have to wait to see which WMG and BMG artists have opted in to be part of those.\n\n## AI music has an ‘overfitting’ challenge\n\nThe addition of a third model to Suno’s family – v6-wild – is very interesting. Why? Because it sheds light on one of the lesser-reported technology challenges facing AI-music companies as they get more popular. It’s an “overfitting” problem.\n\n“Our models are shaped by the preferences of of our users. The more people use our models, the more we understand what people like and what they don’t like – either implicitly or explicitly. That goes back into making the model better,” explained Brody.\n\n“But as you start to orient a model towards what people like on average, sometimes the model actually overfits to those preferences to some degree. On average you’re going to get something that sounds a little bit more poppy because on average people enjoy that.”\n\nWhich is not a problem for those people. But if you’re Suno, and a big part of your outreach to the music industry is trying to persuade professional artists and producers to use your models, it decidedly *is* a problem.\n\n“If you’re an artist and you’re trying to make more of a niche genre, you don’t want it to skew towards pop. So, v6-wild is less tuned towards preferences than v6… Artists are not necessarily looking for the most enjoyable single output. What they’re looking for is a creative partner.”\n\nThis feels like an important insight into a problem that’s going to affect any AI-music model that reaches significant scale, as well as into the solution that Suno has chosen to tackle it.\n\nWe wonder if more generally, the overfitting challenge will influence the licensing strategies of AI-music companies. They can’t just go for the big, mainstream catalogues from the biggest music companies: they’ll need to license weirder, wilder, wonderful independent music too. Although that means persuading weird, wild, wonderful independent musicians to opt in.\n\nThe Believe and TuneCore deal may help with that. “Artists who consent to participate will be compensated for their music and their rights will be protected,” promised Suno and Believe’s announcement.\n\nHowever, this isn’t just about training. “All tracks created by artists using Suno’s new industry partner model will become eligible to distribute through Believe and TuneCore,” noted the announcement. In April this year, Believe banned Suno-made tracks from its pipeline.\n\n## Natural-language editing is key to Suno v6\n\nPart of the launch messaging around v6 is giving more control to musicians as they work with the tracks and sounds they generate using Suno. The big direction of travel here is towards natural-language editing. Think voice prompts rather than sliders.\n\nOr rather: until now, casual users interacted with Suno by typing in text prompts (or even [pasting in screenshots of chats](https://musically.com/2026/06/01/suno-is-riding-the-wave-of-tiktoks-new-text-to-song-trend/)) while the more-hardcore / professional users got stuck in to its Suno Studio DAW – sliders, dials, drop-down menus etc.\n\nv6 goes big on using plain language for some of the deeper tasks within Suno. Example commands in the launch blog post include “Change the chorus so it’s sung by a gospel choir” and “Change the lyric from ‘love’ to ‘light'” – but also “Sample the riff at 0:45, isolate the guitar, build a beat around it” and “Take the vocals from x, drums from y, and add new lyrics about losing control, make it 80s synthwave”.\n\n“It can start to do things that you can describe through natural language that just weren’t possible in the previous generation,” said Brody, who compared this shift – and this feels like a carefully-chosen comparison – to the traditional recording process.\n\n“If you go into a professional studio, you usually have an artist sitting next to a producer, using natural language to describe what they want changed in a song, and the producer has the skills to go and make that happen within the complex interface that they’re using,” he said.\n\nNatural language isn’t always going to be the best way, noted Brody. “Studio 2.0 allows you to get in the weeds and manipulate these things by hand. Oftentimes, that is the fastest or most precise way, and natural language will never be better,” he said.\n\n“It’s very difficult to say ‘Hey, at the third downbeat in the second verse, at 22 seconds, I want it clipped by 4.3…’ You can just move the waveform, right?!… So this is all going in a direction where both of these things [control methods] exist, and they supplement and support each other for different use cases. That’s going to be super-powerful for the professionals as well as the casuals.”\n\n## ‘The foundation for The Everything Model’\n\nSome Music Ally readers will be unhappy with this interview. Why give Suno space to talk about its whizzy new models when it’s still facing all those lawsuits?\n\nFeelings are still running high around this particular company, and AI music generally. We’ve been covering all those lawsuits (and feelings) as comprehensively as we can.\n\nAt the same time… it also feels important to understand – and cover – how Suno is developing as a platform, and how the people building that product are thinking.\n\nBrody certainly has plenty of thoughts that (we think) are a useful addition to the music industry’s knowledge as it seeks to shape an ethical, positive future for this company and this technology.\n\n“I refer to this as the foundation for The Everything Model. And I I do believe that. I think this is the point at which anything you could think to ask the music model to do, it has the technical capability to do,” he said.\n\n“Now, what can get better and what will get better? Everything! The fact that now you can ask it to do everything means that we will have an opportunity to start collecting feedback on all of the things somebody could ask it to do,” added Brody.\n\n“We will start to better understand what are the weaknesses of this model suite. That’s how these things get better over time: you’re constantly getting better on the dimensions that were previously deficient… I think this is the worst the model will ever be. It’s only going to get better from here, and that’s going to come through the hard work of the team and the feedback of the community.”\n\nThen he mentioned the ‘t’ word – taste. Something that many people in the creative industries believe AIs will never have. No lived experience, no soul, no way of truly understanding whether something is good or bad…\n\n‘Only humans have taste’ was a *huge* theme at this summer’s Cannes Lions conference, for example. So it’s interesting – and perhaps challenging – to hear Brody talk about taste in a different way, in relation to the feedback loops from Suno’s users.\n\n“The feedback is the subjective input to taste, and ultimately taste is the thing that starts to orient art towards something that people like or don’t like,” he said.\n\n“That’s the nature of training something that is a kind of non-deterministic, subjective AI model. We’re empowering humans to make art through this model, and so the feedback from those humans matter, and it’s a big part of how we make this thing better over time.”\n\nCan an AI model *really* develop taste though – even if it has tens or hundreds of millions of users giving feedback on everything they generate using that tool?\n\n“If you try different models, you will start to notice different aspects of those models that feel like patterns or feel like taste. Beauty is in the eye of the beholder here, but you can start to tell the differences between these models,” said Brody.\n\n“Some are preferred to others, and that’s taste. it’s taste put into the model – trained into the model – and we’re very intentional with how we do that.”", "url": "https://wpnews.pro/news/suno-rolls-out-v6-song-making-model-will-pay-royalties", "canonical_source": "https://musically.com/2026/09/09/suno-launches-its-v6-ai-music-models-heres-what-you-need-to-know/", "published_at": "2026-09-09 10:53:11+00:00", "updated_at": "2026-09-09 11:13:14.611434+00:00", "lang": "en", "topics": ["artificial-intelligence", "generative-ai", "ai-products", "ai-ethics"], "entities": ["Suno", "Jack Brody", "Warner Music Group", "BMG", "Believe", "UMG", "Sony Music"], "alternates": {"html": "https://wpnews.pro/news/suno-rolls-out-v6-song-making-model-will-pay-royalties", "markdown": "https://wpnews.pro/news/suno-rolls-out-v6-song-making-model-will-pay-royalties.md", "text": "https://wpnews.pro/news/suno-rolls-out-v6-song-making-model-will-pay-royalties.txt", "jsonld": "https://wpnews.pro/news/suno-rolls-out-v6-song-making-model-will-pay-royalties.jsonld"}}