Explainer · Google's music models
Google Lyria 3 Explained
Lyria is Google DeepMind's music-generation model family: Lyria 3 arrived in the Gemini app in February 2026, and Lyria 3 Pro followed on March 25, 2026 alongside a faster Clip variant. Here is what each version does, where Google offers it, and how to try it on this site.
Dates and capabilities below come from Google's own announcements and developer documentation.
The family
What is Google Lyria?
Lyria is a family of AI music models developed by Google DeepMind. Rather than a single model, the family now spans three tiers that trade track length, song-structure control and generation speed. Google says Lyria 3 outputs carry a digital watermark and that the models are designed to avoid mimicking existing artists — useful context if you plan to publish generated audio.
Built by Google DeepMind
Lyria is developed by Google DeepMind and ships inside Google's consumer and developer products rather than as a standalone music website.
Watermarked output
According to Google, Lyria 3 generations carry a digital watermark so AI-produced audio can be identified downstream.
Artist-safe design
Google states the Lyria 3 models are designed to avoid mimicking existing artists.
The three tiers
Lyria 3, Lyria 3 Pro and Lyria 3 Clip
All three belong to the Lyria 3 generation. Here is what Google has documented for each tier so far.
Lyria 3 — February 2026
Google introduced Lyria 3 in the Gemini app in February 2026. It generates tracks of about 30 seconds and is the everyday tier of the family.
Lyria 3 Pro — March 25, 2026
Announced on March 25, 2026, Lyria 3 Pro produces tracks of up to about three minutes and understands song structure — you can prompt intros, verses, choruses and bridges. Google's Gemini API docs describe 48 kHz stereo audio generated from text prompts or image inputs, under model id lyria-3-pro-preview.
Lyria 3 Clip
Lyria 3 Clip (model id lyria-3-clip-preview) is optimized for speed and high volume: roughly 30-second clips aimed at prototyping, loops and social assets.
Official access
Where Google offers Lyria 3
Google distributes the Lyria 3 family through its own products. Depending on tier and region, you can find it in the Gemini app, Google AI Studio, the Gemini API, Vertex AI (in public preview for Lyria 3 Pro) and Google Vids. Access inside Google's products is decided by Google and can change over time.
This site is not Google
lyria3-ai.com (Lyria3 AI) is an independent, third-party website. It is not affiliated with, endorsed by, or operated by Google or Google DeepMind, and nothing on this page is an official Google resource. What this site does offer is a browser-based studio where you can generate music with models named Lyria 3, Lyria 3 Pro and Lyria 3 Clip, using our own credit system.
Try it here
How to try Lyria 3 on Lyria3 AI
Open the studio
Go to the dashboard and pick a model — Lyria 3 is the default, while Lyria 3 Pro and Lyria 3 Clip sit in the same model menu.
Describe the track
Write a prompt for the sound you want, choose vocals or instrumental, and check the credit cost shown before you generate.
Generate and download
Run the generation, listen in the browser, then download the result or iterate with a refined prompt.
Generation on this site uses credits. New accounts get a small number of welcome credits — currently five — so treat your first runs as tests rather than a free library.
Responsible use
A note on publishing generated music
Google states that Lyria 3 outputs carry a digital watermark and are designed to avoid mimicking existing artists. Keep that spirit when you publish: do not pass AI audio off as a specific performer's work, and review the licensing and platform rules wherever you release the track.
Keep reading
FAQ
Google Lyria 3 FAQ
What is Google Lyria 3?
Lyria 3 is a music-generation model developed by Google DeepMind. Google introduced it in the Gemini app in February 2026, where it generates tracks of about 30 seconds from prompts.
Is lyria3-ai.com Google's official site?
No. Lyria3 AI is an independent third-party website and is not affiliated with, endorsed by, or operated by Google or Google DeepMind. We offer a studio that exposes models named Lyria 3, Lyria 3 Pro and Lyria 3 Clip.
What is the difference between Lyria 3 and Lyria 3 Pro?
Per Google, Lyria 3 produces roughly 30-second tracks, while Lyria 3 Pro reaches about three minutes and understands song structure — you can prompt intros, verses, choruses and bridges. Lyria 3 Clip is a speed- and volume-optimized variant for ~30-second clips.
Can I use Lyria 3 in the Gemini app?
Yes — Google introduced Lyria 3 in the Gemini app in February 2026, and offers Lyria 3 Pro there for paid subscribers, plus Google AI Studio, the Gemini API, Vertex AI and Google Vids. Google's own plans decide access inside those products.
Does Lyria 3 generate vocals or lyrics?
Google describes Lyria 3 as able to follow lyrics you provide in a prompt. In our studio you can pick a vocal or instrumental direction; results are AI-generated, so judge quality on your own brief rather than expecting a specific singer's voice.
Is Google Lyria 3 free?
Inside Google's products, availability depends on Google's own plans — we do not state their pricing. On this site, generation uses credits; new accounts receive a small welcome amount (currently five credits) and the exact cost is shown before you generate.
Start generating
Try the Lyria 3 family in your browser
No downloads required. Open the studio, describe a track, and generate with Lyria 3 — Pro and Clip are one dropdown away.