The generative audio landscape is witnessing a massive shakeup. Two of the biggest names in artificial intelligence have launched major updates, aiming directly at the throne currently occupied by market leader Suno.
The Era of Licensed Training Data
Following the high-profile RIAA copyright lawsuits of 2024, “trained on licensed data” has become the most critical phrase for any new launch. Both ElevenLabs and Stability AI are leaning heavily into this, ensuring creators can use these AI music generators without fearing legal repercussions.
“The music industry is no longer fighting generative AI; it is licensing it. The player with the cleanest training data and the most flexible developer tools wins.” — Industry Analyst.
Market Valuations & Financials:
- ElevenLabs: Valued at $11 billion with a $500 million ARR as of April 2026.
- Suno: Valued at $2.45 billion with over $300 million ARR.
ElevenLabs Music v2: Coherence and Price Cuts
Arriving ten months after its predecessor, ElevenLabs’ Music v2 focuses on structural coherence under pressure. The core pitch is impressive: a single track can shift from opera to heavy metal and back, handle rapid-fire rap, and embed non-musical sound effects without the composition falling apart.
The update also introduces highly functional inpainting, allowing users to select and regenerate specific sections of a track while leaving the rest untouched. To drive adoption, ElevenLabs slashed API pricing by up to 50%, making it highly competitive for developers and brands.
Stable Audio 3.0: Open Weights and Local Execution
Stability AI is sticking to its open-source roots with Stable Audio 3.0, releasing a family of four models, three of which feature open weights on Hugging Face:
- Small SFX — On-device sound effects generation.
- Small — Full music composition running locally without a dedicated GPU (459M parameters).
- Medium — Generates tracks up to 6:20 minutes on stronger hardware.
- Large — High-end, API-only model for enterprise clients.
Utilizing a new semantic-acoustic autoencoder architecture called SAME, Stable Audio 3.0 maintains melodic coherence over long durations. It also supports LoRA fine-tuning, enabling artists to adapt the model to their own unique catalogs.
Can They Catch Suno?
Suno remains the undisputed giant, boasting roughly 100 million users and generating 7 million songs daily. However, with ElevenLabs securing licensing deals with Believe, Kobalt, and Merlin, and Stability partnering with Universal Music Group and Warner Music Group, the enterprise market may shift toward these legally compliant alternatives.
Frequently Asked Questions (FAQ)
What makes Stable Audio 3.0 different from other AI music generators?
Stable Audio 3.0 offers open-weight models that can run locally on consumer-grade hardware, meaning developers and creators do not need to rely on cloud APIs to generate music.
Are the tracks generated by ElevenLabs Music v2 safe for commercial use?
Yes. ElevenLabs has established licensing partnerships with major distributors, ensuring that outputs generated on their commercial tiers are fully cleared for professional use.
What is LoRA in AI music generation?
LoRA (Low-Rank Adaptation) is a lightweight model used to fine-tune a larger AI model. In music, it allows creators to train the AI on specific artist catalogs or genres to produce highly customized styles.
