ElevenLabs Releases 'Music v2.5' Commercial Music Model with Free Plan Ownership

Voice and audio AI research firm ElevenLabs announced on September 11, 2026, the official release of 'Music v2.5', a commercial foundation model for music generation designed to deliver acoustic instrument fidelity comparable to live studio performances, alongside sophisticated melodic arrangements. Beyond audio quality enhancements, the release introduces a major restructuring of licensing and distribution terms, granting full ownership rights over generated tracks to all users, including those on the free tier.

Image source: https://elevenmusic.io/blog/introducing-music-v2-5

Historically, generative AI music has faced adoption barriers in commercial production due to synthetic artifacts, insufficient stem separation, and ambiguous licensing terms. With Music v2.5, ElevenLabs aims to address both creative and commercial hurdles by improving structural coherence across complex song sections while establishing transparent, permanent ownership terms for creators.

ElevenMusic Integration with Inpainting and Long-Form Composition in API

Music v2.5 is now deployed as the default engine across ElevenLabs' web-based creation platform, ElevenMusic, powering prompt-based generation and reference-guided workflows.

Creators can generate complete musical compositions from scratch using natural language prompts, isolated audio stems, or rhythm loops. In parallel, advanced editing capabilities powered by Music v2.5 have been rolled out across the ElevenCreative studio suite and the developer-facing Music API.

• Audio Inpainting and Precision Re-arrangement: Creators can selectively isolate specific bars, vocal phrases, or solo instrumental sections to regenerate or re-harmonize individual passages without disturbing the rest of the composition.
• Structured Long-Form Compositions: The model maintains consistent melodic themes and arrangement architecture across standard song structures, seamlessly transitioning through intros, verses, choruses, bridges, and outros.
• Mid-Track Genre Transitions: Compositions can evolve dynamically across genres within a single track, shifting from indie rock into jazz improvisation or blending electronic percussion into full acoustic orchestrations.
• Multilingual Vocal Performance: Pronunciation accuracy, metric stress, and vocal expressive range have seen substantial improvements across multiple languages, facilitating localized vocal production for international creators.

According to ElevenLabs, these architectural refinements allow artists to transform preliminary ideas, ambient loops, or motif sketches into fully orchestrated production-ready tracks.

Commercial Rights for All Tiers and Daily Lossless Audio Downloads

The centerpiece of the Music v2.5 launch is a decisive policy update regarding audio ownership and export fidelity.

ElevenLabs has guaranteed that all users, including those on the Free plan, retain the rights to the music they generate with Music v2.5. Unlike conventional subscription frameworks where commercial rights expire when a paid plan lapses, creators retain ownership of their generated tracks even after downgrading or terminating their subscriptions.

• Free Tier Provisioning: Users on the free plan receive an allowance of 5 lossless audio downloads per day directly from the platform.
• Pro Tier Capacity: For production studios and audio professionals, the Pro subscription tier provides up to 400 lossless downloads per month to support high-throughput commercial asset pipelines.

Creators should note that specific terms governing commercial exploitation and platform distribution remain subject to ElevenLabs' subscription Terms of Service, which vary according to tier tiers and enterprise licensing agreements.

Copyright Protections and Reference Download Safeguards

Alongside expanded commercial permissions, ElevenLabs has implemented rigorous technical guardrails to mitigate copyright infringement and intellectual property conflicts.

When a user provides an existing copyrighted track by another artist as a reference input during generation, the system automatically restricts and blocks lossless audio downloads for the resulting output. This measure is intended to prevent uncredited voice likeness copying, melodic piracy, or unauthorized distribution of derivative material mimicking commercial artists.

ElevenLabs noted that audio watermarking and provenance tracking systems developed for its speech synthesis tools are being integrated into the music pipeline to uphold ethical standards across the emerging generative music ecosystem.

Sources

• ElevenMusic Official Blog: Introducing Music v2.5
• ElevenLabs Official X (@ElevenLabs): 2026-09-11 Music v2.5 Release Announcement