Google announced on March 25, 2026, the release of Lyria 3 and Lyria 3 Pro, its music generation models, now available to developers in public preview through the Gemini API and Google AI Studio, according to blog.google.
According to blog.google, developers can choose between two model variants:
- Lyria 3 Pro (lyria-3-pro-preview): Creates tracks up to approximately three minutes long with “professional-grade structural awareness” and “studio-quality, premium output.”
- Lyria 3 Clip (lyria-3-clip-preview): Optimized for speed and high-volume requests.
According to TechCrunch, Lyria 3 Pro allows users to specify different elements in prompts such as intros, verses, choruses, and bridges, as the model “understands track structure better than its predecessor.” This contrasts with the original Lyria 3 model, which offered 30-second tracks, TechCrunch reports.
The Pro model is rolling out in the Gemini app for paid subscribers only, according to TechCrunch. Google is also integrating Lyria 3 Pro into Google Vids and ProducerAI, a music production tool Google acquired last month, TechCrunch notes.
According to TechCrunch, Google stated it trained the model using “data from its partners and permissible data from YouTube and Google,” and emphasized the model doesn’t mimic artists. All generated tracks are marked with SynthID to denote AI creation, TechCrunch reports.