Suno introduced Suno v6, a new model family it says was trained on licensed data from music labels and distributors including Warner Music Group, BMG and Believe. The company released three versions: the base Suno v6 model, available to paying users and described as reliable and steerable for controlled outputs; Suno v6 wild, an experimental model for paying users meant for ideation and unexpected results; and Suno v6 mini, a faster version available to all users. Suno plans to retire its older models and said v6 was not trained on the data used to train previous versions of its music-generating model.

The release comes as Suno faces lawsuits from record labels that accused the startup of using copyrighted data to train its AI models. Suno settled with Warner Music Group last year and struck a deal with BMG last month. It is still fighting litigation from labels including Sony and Universal Music Group, artist Jason Isbell, and users who allege the company ignored security while chasing profits. The announcement also follows Suno’s admission that it trained models using YouTube videos.

Suno v6 adds more flexible creation tools. Users can edit part of a song using a prompt or a word in the lyrics, use text, images, or video as references to create tracks, and separate an instrument from a sample to create a new beat. Suno also plans to add remixing features, provided artists opt in to a new program it is setting up with music labels. Chief product officer Jack Brody said derivative work creates more revenue opportunities for rights holders and artists, and that Suno can help manage ecosystem issues like streaming fraud and mass exportation of low-intent uploads, while governance should remain with distributors and platforms.

Suno previously said it would add watermarking to songs generated on its platform and introduced download limits based on account tier. Despite ongoing legal pressure, Suno has raised over $819 million in funding to date, according to PitchBook data.