Suno's v6 AI music model trained on licensed data from Warner Music Group

Suno's v6 marks a shift in how the company builds its music generation tools, moving to a model trained from the ground up on a new data set that does not include the same data as previous models. The inclusion of licensed content from major players like Warner Music Group, BMG, and Believe suggests a closer alignment with the record industry, though it remains unclear whether the training data is completely free of dubiously obtained content. This release also introduces a tiered approach with three distinct models—v6, v6-wild, and v6-mini—catering to different needs from free, fast creation to more unpredictable outputs.

The updates extend beyond the model itself, with new editing capabilities that let users modify parts of a song using plain language within a chat, and the ability to mash together elements from their Suno library. Prompts can now be based on images, video, or other audio, not just text. These changes, alongside the eventual retirement of previous models, point to a platform evolving to give users more granular control. However, the model still struggles with imperfections; attempts to generate off-key or out-of-tune results failed, and it remains incapable of harmonic and rhythmic imperfection. Whether this limitation matters to users will depend on how much they value natural imperfections over the polished output Suno's v6 delivers.