Google put Lyria 3.5 in the Gemini app and opened the API to developers

Google made Lyria 3.5 available in the Gemini app and the Gemini API on September 4, 2026. The model itself is not new. It shipped in Google Flow Music on July 29, and for five weeks that was the only place you could use it. Now it sits inside the assistant Google puts in front of a general audience, and behind a public API.

That is the whole story, and it matters more than a version bump would. A model most producers had to seek out is now a default option in a chat box.

What changed in the Gemini app

Google lists three things. You can select or describe your genre and choose between vocal or instrumental styles. You get templates as starting points, which Google pitches for background music, brand jingles, ringtones and custom birthday tracks. And you can pick a short or a longer track.

The genre and vocal controls are the useful pair. Describing a style in plain words and toggling vocals off is the minimum you need before a generated bed is worth putting under anything.

“

Lyria 3.5, our best-sounding music generation model, is now available in the Gemini app and the Gemini API.

The template framing tells you who Google is aiming at. Birthday tracks and ringtones are not producer features. They are prompts for people who have never opened a DAW, a piece of software like Ableton or Logic where music actually gets built.

Why the Gemini API release is the bigger half

Putting Lyria 3.5 behind the Gemini API means any developer can call song generation from their own product. Google also lists Google AI Studio and Google Vids as access points.

Compare that to how Suno and Udio reach people. Those are destinations. You go to the site, you generate, you leave. An API turns generation into plumbing that shows up inside apps that have nothing to do with music.

The catch: Google is still fighting an indie-artist lawsuit over how the Lyria family was trained, and I covered Google’s YouTube terms defense in June. Wider distribution does not settle the training question. It raises the stakes on it.

Where this leaves Flow Music

Google Flow Music keeps the model too, and Google still frames it as the home for artists and AI creatives. My coverage of the July Flow Music launch has the detail on tempo, drums, bass and track-length controls, which are still the deeper control set.

So the split is roughly: Gemini for a quick track, Flow Music when you need to steer the parts. Android Authority notes the Gemini version leads with templates and length choice rather than the full control surface.

Every output still carries SynthID, Google’s embedded watermark for AI-generated content. That does not change with the surface you generate on.