Suno released v6, its first AI music model built using data licensed from Warner Music Group, BMG, a...
The AMW Read
Converts previously announced label licensing deals into a shipped product, advancing the segment's IP-licensing frame without fully resolving training-data provenance questions.
Suno released v6, its first AI music model built using data licensed from Warner Music Group, BMG, and Believe alongside user-generated content, the company said. The launch ships as three variants: v6 for general use, v6-wild for looser and more unpredictable output, and v6-mini, a free, resource-light tier. New features let users edit song elements — a guitar line, a lyric — through chat without regenerating the full track, and accept image, video, or audio prompts alongside text. Suno said it will eventually retire its earlier models. Genre adherence improved noticeably over prior versions, though the model still cannot reliably produce the off-key or dissonant imperfections of real human performance.
This follows Suno's licensing agreements with Warner Music Group and BMG announced a day earlier, converting those catalog deals into a shipped product for the first time. Suno, tracked in the AI Market Watch index with $775M in total funding since its 2022 founding (per the AI Market Watch index; coverage, not a census), had operated for years under active copyright disputes with major labels over training data; v6 is the company's attempt to demonstrate a licensed-data path forward. The Verge's testing found it unclear whether v6's training set is fully free of unlicensed material, since Suno has not disclosed the deal's full data scope.
For builders, the shift signals that label-negotiated revenue-sharing is becoming a cost of doing business for AI music generators, raising the bar for smaller entrants without label relationships. Investors should watch whether v6's editing and multimodal-prompt features — not just the licensing headline — become the actual retention driver, since raw text-to-audio generation is becoming table stakes across competitors.
