Two major AI music updates arrived this week, and neither came from Suno.
ElevenLabs, the Polish-founded voice AI company carrying an $11 billion valuation following a $500 million Series D in February, has launched Music v2. Meanwhile, Stability AI — the company behind Stable Diffusion — has released Stable Audio 3.0, a four-model family featuring open weights and tracks that extend past six minutes.
The backdrop to both announcements is the Recording Industry Association of America's 2024 copyright lawsuits against Suno and Udio, which made "trained on licensed data" the most consequential phrase in any AI music release. Both ElevenLabs and Stability AI are leaning heavily on that point, signaling that outputs generated by their tools should be free of legal complications for users.
Music v2: One Track, Opera to Heavy Metal, No Breakdown
Music v2 is ElevenLabs' second music model, arriving roughly ten months after the first. The core pitch is coherence under pressure. According to ElevenLabs, a single track can shift from opera to heavy metal and back, hold together through fast rap, and embed non-musical sound effects — all without the composition falling apart. Generative audio tends to degrade exactly when prompts get complicated, making this capability particularly significant for longer compositions.
Inpainting is now a practical feature: users can select a section, regenerate it, and leave everything else untouched. Songs can also be built section by section — intro, verse, chorus — with the model maintaining continuity throughout rather than treating each clip as a standalone generation. Multilingual support has also improved, though ElevenLabs did not publish specific details.
The model powers three platforms: ElevenMusic for creators, ElevenAPI for developers, and ElevenCreative for brands. ElevenLabs also cut Music v1 and v2 pricing by up to 50% for ElevenAPI and up to 40% for ElevenCreative self-serve.
The company reached $500 million in annual recurring revenue in April 2026. Music remains a small portion of that figure, but ElevenMusic — which launched as a consumer app in April — is a direct challenge to Suno's user base.
Stable Audio 3.0: Open Weights, On-Device, and Longer Tracks
Stable Audio 2.0 topped out at three minutes and was already trailing Suno when it launched in 2024. Stable Audio 3.0 ships four models: Small SFX (on-device sound effects), Small (full music composition on-device), Medium (up to 6 minutes and 20 seconds, requiring stronger hardware), and Large (API-only).
Three of the four variants have open weights available on Hugging Face. The Small models run at 459 million parameters each and require no GPU. Medium hits 1.4 billion parameters and generates its 6:20 output in approximately 1.31 seconds on an H200 GPU. Large, at 2.7 billion parameters, is API-only and targeted at organizations with over $1 million in revenue.
Per-second generation granularity means users receive exactly the track length they requested, not an approximation. The suite is also supported in ComfyUI for local setups.
The underlying architecture is new: a semantic-acoustic autoencoder Stability calls SAME, designed to maintain melodic coherence over longer outputs. LoRA fine-tuning is supported, allowing artists to adapt the models to their own catalogs. A LoRA (Low-Rank Adaptation model) functions like a small conditioning layer on top of the full model — train one on blues, and the model produces blues; train one on a specific artist's style, and the outputs will reflect that sound.
Inpainting is also included, covering single-segment, multi-segment, and causal continuation to extend a track past its original endpoint. Inpainting allows a model to correct specific errors within a generated piece — for instance, selecting a few seconds at the 2:30 mark, prompting a replacement, and having the model generate audio that blends seamlessly with the surrounding track.
Stability has been technically credible in AI music for years without achieving significant commercial breakthrough. The open-weight approach mirrors the Stable Diffusion strategy applied to audio: seed the developer community and see what gets built. Licensing is cleaner than anything Stable Audio has previously shipped, with partnerships in place with Universal Music Group and Warner Music Group.
The Target: Suno, the AI Music Leader
If ChatGPT is the dominant force in AI text, Suno holds that position in AI music. The company reached a $2.45 billion valuation in November 2025, crossed $300 million in annual recurring revenue, and has been used by roughly 100 million people. It generates around 7 million songs per day.
Warner Music settled its lawsuit against Suno in November 2025; Sony and Universal Music Group remain in federal court. To sidestep similar legal exposure, ElevenLabs has secured licensing deals with Believe, Kobalt, and Merlin. Stability has agreements with Warner and Universal. Udio settled with all three major labels and now operates as a walled garden — nothing generated on the platform can be exported.
Stable Audio 3.0 Small and Medium are available on Hugging Face now. Music v2 is free for ElevenMusic users, with commercial tiers available through ElevenCreative and ElevenAPI.
Why it matters
The RIAA's 2024 lawsuits against Suno and Udio have made licensing provenance a structural requirement for commercial AI music tools — both new releases explicitly address this, which affects whether businesses can use outputs in commercial projects without separate legal review.
Udio's post-settlement restriction — no track exports — illustrates a concrete risk for users who build workflows around platforms that later settle litigation on restrictive terms; the open-weight approach of Stable Audio 3.0 offers a different risk profile since the models can be run locally.
LoRA fine-tuning support in Stable Audio 3.0 means artists and studios can adapt the base model to a specific sound catalog without retraining the full model, a capability with direct implications for music production houses managing proprietary style libraries.
