Suno has officially unveiled its v6 family of AI music generation models, signaling a significant shift in the company’s approach to artificial intelligence and the music industry. For the first time, Suno collaborated directly with major record labels, including Warner Music and BMG, to train the new systems. The v6 lineup includes three distinct models: the flagship v6, the experimental v6-wild, and the accessible v6-mini.
Jack Brody, Suno’s chief product officer, described the flagship v6 model as reliable and precise, designed to deliver polished musical output across diverse genres. Meanwhile, v6-wild caters to users seeking less predictable results, serving as a creative brainstorming tool. “Musicians and producers often aren’t looking for the single best output,” Brody explained. “They’re looking for a brainstorming partner.”
While v6 and v6-wild are restricted to Pro and Premier subscribers, v6-mini remains available to free users. The new models boast improved speed, generating tracks within seconds of a prompt, and introduce novel features such as natural language editing of specific song sections and the ability to create mashups from multiple source audio files. Additionally, multi-modal capabilities now allow the system to generate songs from images, video, or emotional descriptors, and users can replace individual lyrics without regenerating entire tracks.
This launch represents the formal commencement of Suno’s partnerships with major labels, following a turbulent legal history. In 2024, a coalition of major record labels, led by Warner Music, sued Suno and its competitor Udio for alleged massive-scale copyright infringement. The conflict shifted in 2025 when Warner dropped its lawsuit against Suno in exchange for a licensing agreement intended to give artists greater control over how their voices and music are utilized in AI-generated content.
Brody stated that the partnership begins generating revenue for partners immediately upon the September 9 launch, creating new avenues for artists to remix and monetize their work. However, questions regarding the legitimacy of Suno’s training methods persist. A 2026 leak revealed that the company had previously scraped decades of music and podcasts from platforms like YouTube and Deezer to train earlier models. As part of the v6 rollout, Suno is retiring those older models to replace them with the newly trained data sets.
The ability to swap out lyrics without regenerating the whole track is a game changer for songwriters experimenting with hooks.
Great to see labels partnering instead of suing, but I still worry about what the v6 models are actually trained on.