How long a song can Lyria 3.5 write?

Lyria 3.5 produces tracks of up to about three minutes that contain verses, choruses and bridges, and Google's API docs say the duration is steered through the prompt rather than a parameter. A second model id, lyria-3-clip-preview, always returns a fixed 30-second clip instead, which is the cheaper option for stings and loops.

Where can I use Lyria 3.5?

Google says Lyria 3.5 is open to all users globally on the web at gemini.google and in the Gemini mobile app. Artists and AI creatives reach it in Google Flow Music, developers reach it through Google AI Studio and the Gemini API, and it also shows up in Google Vids inside Google Workspace.

Can I edit a track after Lyria 3.5 generates it?

No. Google's documentation describes generation with Lyria 3.5 as a single-turn process with no iterative editing, so a change means sending a fresh request. The same docs warn that identical prompts can return different results, and that Lyria 3.5 rewrites the prompt internally without showing you the rewritten version.

Is Lyria 3.5 audio watermarked?

Yes. Google states that every clip Lyria 3.5 generates carries a SynthID audio watermark, which Google describes as imperceptible to the human ear, so a file can be identified as AI-generated later. Google also runs safety filters over every prompt before the model generates anything.

Google · 2026-09-04 · major

Lyria 3.5 comes to Gemini — Google's music model lands in the app and API

Lyria 3.5 is Google's music generation model, and it is now in the Gemini app for every user worldwide as well as in the Gemini API. Developers call it as the model id lyria-3.5 and get 44.1 kHz stereo tracks of up to about three minutes.

Google Lyria 3.5 announcement graphic for music generation in the Gemini app

Google's music model is now one API call away, and free for every Gemini user.

Quick facts

Maker	Google DeepMind
Model ID	lyria-3.5
Audio	44.1 kHz stereo, MP3 by default (WAV optional)
Track length	Up to about 3 minutes
Clip model	lyria-3-clip-preview, fixed 30 seconds
Availability	Gemini app, Gemini API, Google AI Studio, Flow Music, Google Vids
Watermark	SynthID on all generated audio

What is it?

Lyria 3.5 now writes music inside the Gemini app for everyone worldwide, and Google has opened the same model to developers through the Gemini API and Google AI Studio. In the app you name or describe a genre, pick vocal or instrumental, and can start from templates covering everything from background music to a personalised birthday song. Google calls it its best-sounding music generation model.

How does it work?

Two model ids sit behind the feature. The full model, lyria-3.5, writes complete songs with verses, choruses and bridges; lyria-3-clip-preview always returns a fixed 30-second clip. A request can carry plain text, custom lyrics with section tags and instrument timestamps, or up to 10 images. Output is 44.1 kHz stereo, MP3 by default with WAV available on the full model, and the docs note that Lyria 3.5 rewrites prompts internally without exposing the rewrite.

Why does it matter?

Putting a music model behind a Gemini API model id means a developer building a video editor, a game or a content pipeline can request a backing track in the same flow that already calls Gemini for text, instead of wiring in a separate music vendor. On the consumer side, the Gemini app rollout hands full-length song generation to every user globally rather than to the smaller Flow Music audience it launched with in July. The SynthID watermark travels with each file, so the output stays identifiable downstream.

Who is it for?

developers building media, video and game tools

Frequently asked questions

How long a song can Lyria 3.5 write?: Lyria 3.5 produces tracks of up to about three minutes that contain verses, choruses and bridges, and Google's API docs say the duration is steered through the prompt rather than a parameter. A second model id, lyria-3-clip-preview, always returns a fixed 30-second clip instead, which is the cheaper option for stings and loops.
Where can I use Lyria 3.5?: Google says Lyria 3.5 is open to all users globally on the web at gemini.google and in the Gemini mobile app. Artists and AI creatives reach it in Google Flow Music, developers reach it through Google AI Studio and the Gemini API, and it also shows up in Google Vids inside Google Workspace.
Can I edit a track after Lyria 3.5 generates it?: No. Google's documentation describes generation with Lyria 3.5 as a single-turn process with no iterative editing, so a change means sending a fresh request. The same docs warn that identical prompts can return different results, and that Lyria 3.5 rewrites the prompt internally without showing you the rewritten version.
Is Lyria 3.5 audio watermarked?: Yes. Google states that every clip Lyria 3.5 generates carries a SynthID audio watermark, which Google describes as imperceptible to the human ear, so a file can be identified as AI-generated later. Google also runs safety filters over every prompt before the model generates anything.

Try it

Call model id lyria-3.5 in the Gemini API, or open gemini.google