Suno releases v3
The company called it its first model able to produce radio-quality output, letting free users generate two-minute songs and paying users four-minute tracks.
- Models & capabilities
- Minor
Suno released version 3 of its text-to-music model to all users, three months after opening its public web app. The company described it as the first Suno model capable of producing radio-quality music, a step up in audio fidelity from the versions that had powered the app’s initial public launch.
The release expanded the range of musical styles and genres the system could produce and improved its interpretation of text prompts, which Suno said resulted in fewer hallucinated lyrics and more natural song endings — both had been visible weaknesses in earlier versions. Free-tier users could generate full two-minute songs, while paying Pro and Premier subscribers, who had tested the model during an alpha period, could generate four-minute tracks. The model supported multiple languages.
Suno did not publish technical details of the underlying architecture or training data, consistent with its approach throughout the period; the company was already facing scrutiny over the source of its training material, which record labels later cited in a 2024 copyright lawsuit. V3’s jump in perceived audio quality nonetheless marked a widely noted point at which AI-generated music began to sound convincingly close to produced tracks rather than an obviously synthetic novelty, intensifying both consumer interest and industry concern over generative music tools.