ElevenLabs brings Music v2 to its API with chunk-based composition

ElevenLabs has moved Music v2 into its developer API. The June 15 changelog says developers can set model_id to music_v2 on Generate music, Stream music, Generate music detailed, and Upload music, turning the model from a creator-facing release into something product teams can wire into automated audio workflows.

That lifecycle distinction matters. ElevenLabs had already announced Music v2 as the model behind ElevenMusic, ElevenCreative, and future ElevenAPI access. The new developer signal is API availability, plus a different control surface: chunk-based composition plans instead of the simpler prompt-shaped music_v1 flow.

Music v2 now has an API path

The changelog lists music_v2 support across generation, streaming, detailed generation, and upload endpoints. It also adds a v2 composition-plan shape built from GenerationChunk and AudioRefChunk segments, so developers can define structure, pacing, references, and arrangement more deliberately.

That is useful for products that need repeatable output, not just one-off audio experiments. A media tool can generate an intro, verse-like bed, ad-safe bridge, and outro as separate chunks, then keep those sections coherent through one plan. Upload flows can also accept the new plan shape, which gives teams a migration path rather than a single breaking rewrite.

Chunk plans change how teams prompt music

Most AI music interfaces ask for a single descriptive prompt. Music v2 still supports descriptive control, but the API update moves serious usage closer to production composition: generate a plan, adjust its chunks, reference existing audio where allowed, then call the endpoint with a structured request.

For developers, that means prompt quality is no longer the only lever. Applications can expose a timeline-like interface, generate multiple variations for a specific section, or keep brand audio constraints consistent across many outputs. The API docs also say request and response bodies accept either v1 or v2 plan shapes, which lets teams migrate endpoint by endpoint.

ElevenLabs ties the release to licensed training data

ElevenLabs positions Music v2 around commercial use, saying the model is trained on licensed data and that generated tracks are cleared for commercial deployment under its terms. That does not remove every music-law question. Teams still need to avoid prompts that intentionally imitate protected works, preserve generation records for client projects, and check platform-specific policies for broadcast, film, or large campaign use.

The independent MindStudio guide frames the same practical point: the main business value is not simply making background music, but generating licensed audio inside repeatable content pipelines. For agencies, app builders, and video teams, API availability is the difference between manual asset creation and a scalable production step.

Pricing and migration notes

ElevenLabs says Music v1 and Music v2 pricing was cut by up to 50% for ElevenAPI and up to 40% for ElevenCreative self-serve customers around the Music v2 rollout. Check current plan limits before committing volume, because music generation costs can move quickly once it sits inside automated ad, podcast, game, or short-video workflows.

Teams already using music_v1 should audit where they rely on MusicPrompt, because the changelog states that MusicPrompt remains available for music_v1 only. The cleaner migration path is to add music_v2 to one endpoint first, inspect the generated composition plans, then decide whether structured chunks produce enough quality or control improvement for the workflow.

For adjacent workflow ideas, LinkLoot’s AI workflow automation guide is a useful companion: treat music generation as one production node beside script, voice, image, and video generation, not as an isolated toy.

Sources and methodology

The primary source is ElevenLabs’ June 15 API changelog because it confirms music_v2 availability and the exact endpoint/model-id changes. ElevenLabs’ own Music v2 launch page provides capability, pricing, and licensing context. MindStudio’s guide was used as independent workflow context, not as the source of the API availability claim.