Suno v6 Operational

Suno v6 is the next-generation model family launched by Suno, the AI music generation platform. It includes the flagship v6 model (for Pro/Premier users), the experimental v6-wild, and the free v6-mini. Compared with the previous generation, the models generate much faster, support natural-language editing of specific song sections, mix multiple audio sources, and extract samples. They can also generate songs from audio and images. Suno also announced partnerships with Warner Music and BMG, offering tiered model packages for different creator groups.

Rapid generation: Enter a prompt and receive a complete song within seconds, with generation speeds far ahead of the previous generation.
Natural-language local editing: Use a prompt or specify a word in the lyrics to regenerate only a particular section of a song.
Multimodal reference generation: Use text, images, or even video as reference material for creating new songs.
Sample extraction: Isolate a single instrument from a song and build a new beat around it.
Licensed Remix (planned): An artist opt-in remix program will allow fans to remix artists' work while the artists receive a share of the revenue.
Tiered models: v6 (flagship) and v6-wild (experimental) are available to Pro/Premier subscribers, while v6-mini is free for everyone.

Training from scratch on licensed data: The model is a completely new model trained from zero. Its training data comes exclusively from catalogs licensed by partners such as Warner Music, BMG, and Believe, and does not include data from Universal or Sony. Partner record companies can choose whether to participate in training and receive corresponding compensation.
Preference-feedback optimization: Suno uses an approach similar to RLHF. Each song generates two versions by default, and the model continuously optimizes its preferences based on which version users keep, gradually aligning its output with popular listening preferences.
Multimodal conditional generation architecture: The model encodes heterogeneous inputs such as text, images, and video into unified conditioning signals that guide the audio generation process, enabling cross-modal creation such as generating a song from a single image.
Local regeneration (inpainting): Borrowing from image restoration, v6 can mask and locate specific time segments of a song and regenerate audio only within the selected section while keeping the rest unchanged. This supports natural-language editing and word-level lyric modifications.
Source separation and sample extraction: Built-in stem separation can split a mix into independent tracks such as vocals, drums, and bass. It supports accurately extracting individual instrument samples from tracks for secondary creation, with cleaner separation and fewer artifacts than the previous generation.

Create an account: Visit the Suno website to register and sign in. Free users can use v6-mini, while Pro/Premier subscriptions unlock the flagship v6 and experimental v6-wild models.
Generate with a prompt: Describe the song's style or theme in the creation box, or paste lyrics, then click Generate to receive a finished song within seconds.
Choose a version: Two versions are generated by default for each song. Preview them and keep the one you prefer; your choice helps optimize the model.
Edit a section: For an unsatisfactory passage, use a natural-language instruction or select a word in the lyrics to regenerate only that part.
Create with multimodal references: Upload an image or video as a reference and have the model generate accompanying music or a song.
Extract samples: Use track separation to isolate an instrument and build a new beat or remix around it.

Rapid generation: A finished song can be produced within seconds to about 30 seconds after entering a prompt, delivering significantly higher production efficiency than comparable tools.
High-quality full-song completion: Generate a complete song up to 4 minutes long in one pass, with coherent structure and well-developed lyrical rhythm, without repeated extension and stitching.
Precise control: Natural-language local editing, word-level lyric changes, and sample extraction provide industry-leading creative control.
Fully licensed training: Trained from scratch on catalogs licensed by Warner Music, BMG, and Believe, with far lower legal risk than competitors.
Open ecosystem: Download works and distribute them to third-party platforms such as Spotify through Believe/TuneCore.
Tiered model suite: The flagship v6, experimental v6-wild, and free v6-mini cover everyone from professional producers to casual enthusiasts.

Official website: https://suno.com/blog/introducing-v6

Demo creation for independent musicians: Songwriters enter lyrics and a style description to produce a complete, commercially distributable demo within seconds, greatly shortening the creative validation cycle.
Music for short videos and social media: Creators upload a video or image as a reference to generate music that matches the mood of the visuals while avoiding copyright risks.
Advertising and brand music: Marketing teams generate multiple versions of advertising songs tailored to a product's identity, then release and deploy them directly through Believe/TuneCore.
Fan remix engagement: With the upcoming artist-authorized Remix feature, fans can create remixes around an idol's work and receive a share of the revenue, activating the fan economy.
Music education and inspiration: Music enthusiasts and beginners can explore unfamiliar genres with v6-wild or refine their own work through natural-language editing, lowering the barrier to music creation.