Stability AI just dropped Stable Audio 3.0, a new family of music generation models with a flagship that can produce professional-grade tracks running over six minutes. According to TechCrunch AI, that’s more than double the length the company’s previous version could handle, and a massive leap over the 47-second clips its open-weight release squeezed out last year. The launch puts Stability back in the music conversation right as Google, ElevenLabs, Suno, and Udio are all fighting for the same turf.
What shipped
Stability AI is rolling out four models under the Stable Audio 3.0 banner, each tuned for a different use case:
- Small SFX (459M parameters): Built for sound effects, light enough to run on-device for clips up to two minutes.
- Small (459M parameters): Music generation up to two minutes, also designed for on-device use.
- Medium (1.4B parameters): Full compositions of 6 minutes, 20 seconds, with maintained musical structure and melodic tone.
- Large (2.7B parameters): The flagship, same length ceiling as medium but with more horsepower for professional output.
Who gets what
Three of the four models ship with open weights. TechCrunch AI reports that small SFX, small, and medium are all available for anyone to download and modify. The large model stays locked behind Stability’s API and self-hosting paid services, and any company pulling more than $1 million in revenue needs an enterprise license to use it.
The licensing angle matters
This is where the story gets interesting. Suno and Udio are tangled in court fights with major labels over training data, and that legal pressure is shaping who survives in the AI music space. Stability went the other direction. The company inked deals with Warner Music Group and Universal Music Group last year, and says Stable Audio 3.0 is built entirely on fully licensed data. That’s a structural advantage if the courts force the industry toward licensed-only training sets.
A bet on professional musicians
Stability also signaled it’s building a new product suite aimed at professional musicians, though it kept the details quiet. To lead that effort, the company brought in Ethan Kaplan, former chief digital officer at Universal Audio and Fender. Kaplan’s hire fits a broader pattern across the industry. Suno recently brought on former Merlin CEO Jeremy Sirota as chief commercial officer. ElevenLabs picked up Derek Cournoyer from Kobalt to lead strategy for its music push. Every serious player is stacking up music industry credentials, which says a lot about where this market is heading.
Why this matters
Two things stand out. First, six minutes is the threshold where AI-generated audio stops being a novelty and starts being usable for actual songs, podcasts, game soundtracks, and ad work. Second, Stability’s licensed-data position could turn into a real moat if regulators or courts crack down on competitors trained on scraped catalogs. The race isn’t just about model quality anymore. It’s about who can ship at scale without getting sued into the ground.
The full breakdown is available at the original TechCrunch AI report.