On September 10, 2026, Suno launched the v6 family of music-generation models, in a move the company developed in cooperation with music-industry entities, including Warner Music Group, BMG, and Believe. The family consists of three models: v6 for more controlled and predictable production, v6-Wild for experimentation and unexpected results, and the free, lighter-weight v6-mini.
The launch represents an important shift in Suno’s trajectory. Like other companies in the music-generation field, Suno has faced tensions with the music industry over copyright and the data used to train models. According to the company, v6 was trained from scratch using a new dataset that includes licensed content from its partners, along with data provided by users. Suno says that the training data used in its earlier models was not included in training the new model, but it has not disclosed the full scope of the new dataset.
Three models for different uses
v6 is aimed at users who want results that are more consistent with a specific creative goal, while v6-Wild was designed to push the model toward less predictable ideas and results. Both models are available to Pro and Premier subscribers. v6-mini, meanwhile, is available to all users for free, and Suno is betting on its speed compared with other free music-generation models.
Tests reported by The Verge showed that v6-mini’s results may be simpler than those of the other two versions, with audio distortions associated with AI generation appearing more frequently. By contrast, the v6 models appeared to have improved in understanding musical genres and descriptions of styles; v6 interpreted prompts related to hyperpop and krautrock more accurately than previous models. The models still struggle to imitate some deliberate imperfections in human performance, such as off-key singing or instruments that are out of sync.
Editing parts of a song instead of regenerating it
The update is not limited to the generation model itself. Through conversation and natural language, users can modify a specific part of a song, such as the lyrics of a section or a guitar phrase, without recreating the entire work. Suno also allows users to combine different outputs already in their library and create music from images, video clips, or audio files, rather than from text prompts alone.
The company plans to transition its platform gradually to the v6 family while phasing out older models. This means that the differences among the three models, particularly the audio limitations in the free version, will become important practical factors for users when choosing a subscription tier or workflow.
What does this mean for the relationship between AI and artists?
Suno’s current partnerships go beyond the issue of providing training data. The company is working on AI products centered on specific artists, with an optional participation arrangement in which the artist decides whether to take part. According to what CEO Mikey Shulman told The Verge, artists will receive revenue if they participate in these experiments, which could allow fans to create new content based on a particular artist’s music within a licensed, revenue-sharing framework.
From certi.news’s perspective, the most important change is not merely an increase in the model’s release version or the addition of editing tools, but an attempt to shift the relationship between AI companies and the music industry from copyright disputes toward licensing and revenue sharing. However, the source does not yet explain how artists’ revenue will be calculated, the mechanism for granting and withdrawing consent, or the limits on using their works in future products. The details of the new training dataset have also not been fully published, leaving the assessment of the legal and technical model open.