Google’s new Lyria 3 Pro stretches AI music to three minutes—just long enough to outlast most TikToks.
The search giant is slipping its most advanced music model into Google Vids, Gemini, Vertex AI, Google AI Studio and a new web tool called ProducerAI. The pitch: prompt for intros, verses, choruses and bridges, then let the model stitch them into a coherent track up to 180 seconds long.
Vertex AI customers get first crack at the Pro tier, now in public preview, for what Google calls “on-demand audio at scale.” Studios can batch-generate soundtracks for games or platforms without leaving Google’s cloud console.
Developers using the Gemini API or Google AI Studio can toggle between Lyria 3 and the real-time variant, trading speed for structure. Inside the consumer Gemini app, longer generations are live for paid subscribers this week.
Google Vids, the company’s AI-powered video creator, adds both Lyria 3 and Lyria 3 Pro so Workspace and AI Pro & Ultra subscribers can score marketing clips without licensing headaches. ProducerAI, a browser-based workspace Google built “by musicians,” offers the same agentic loop to free and paid users worldwide.
All outputs are watermarked with SynthID, Google’s inaudible fingerprint, and the company repeats its usual guardrails: no artist mimicry, no IP rip-offs, prompt filters and Terms-of-Service enforcement. Grammy-winning producer Yung Spielburg already used an earlier Lyria build for the Google DeepMind short “Dear Upstairs Neighbors,” and François K has a Lyria-assisted single in the pipeline.
Bottom line: Google just turned its music AI into a multi-app utility. If you wanted royalty-free, three-minute songs on demand, the buffet is open—so long as you stay inside Google’s ecosystem.