Suno Launches v6 Music Models with Licensed Training Data from Warner and BMG

Suno has launched its v6 generation of AI music models, developed in partnership with Warner Music Group, BMG, and Believe. [1][3] Chief product officer Jack Brody stated that the models were trained using licensed Warner music alongside Suno user data. [3]

Suno plans to pay royalties to labels and publishers when the models are used, and the agreements include revenue sharing with rightsholders. [2][3] The company is moving its product entirely to the new generation and will retire its previous models. [1][3]

The v6 generation introduces a suite of three distinct models, departing from Suno's previous single-model approach. [1] The flagship v6 model and the less predictable v6-wild model are available to paid subscribers, while the v6-mini model is accessible to free users. [1][3] Users can now utilize text, audio, images, and video as inputs. [3]

Creators can also use natural language prompts to edit specific parts of existing songs, such as altering a single lyric or chorus. [1][3] Suno continues to block users from directly prompting the system to generate music in the likeness of known artists. [3]

Suno's pivot to licensed training data with major partners like Warner Music Group and BMG marks a significant shift from the scraping practices that previously triggered industry backlash. By integrating revenue sharing and royalty payments, Suno is attempting to align its commercial incentives directly with rightsholders. The introduction of three distinct models mirrors the tiered offerings seen from major AI labs, tailoring specific compute costs and capabilities to free users versus paying professionals. What the sources don't address: Whether the revenue sharing agreements cover all tracks generated by users or only those incorporating specific label-owned elements.