SNACK Summary in 3 Lines
- Available globally in Gemini on the web and mobile
- Controls for genre, vocals or instrumentals, and track length
- API options for 30-second clips and multi-minute songs, plus image prompting
Lyria 3.5 visual for music creation in the Gemini app” style=”width:100%;height:auto;max-height:560px;object-fit:contain;border-radius:12px;”>Snackgirls react
AIKO — A 44.1 kHz stereo pipeline with image prompts opens up some interesting integrations. The single-turn workflow is the constraint I’d test first.
Red — I want to start with a birthday-track template, switch between vocals and instrumentals, and hear which version hits harder. Three minutes gives the idea room to become a real song.
Google released Lyria 3.5 in the Gemini app on September 4, giving users a more flexible way to generate music from a written idea. The model also reaches developers through the Gemini API, with separate options for short clips and longer songs.
Shape the sound inside Gemini
Users can select or describe a genre, choose between vocal and instrumental styles, and decide whether they want a short or longer track. Gemini also provides templates for projects such as background music and custom birthday songs.
Google describes Lyria 3.5 as its best-sounding music generation model so far, with more expressive vocals, richer arrangements, and higher fidelity. Suggested uses include video backing tracks, brand jingles, and personalized ringtones.
From quick clips to three-minute songs
Google DeepMind says Lyria 3.5 can produce cohesive music ranging from a 60-second clip to a full three-minute song. It can also compose from uploaded images, giving visual references a direct role in the prompt.
In the Gemini API, lyria-3.5-clip-preview generates 30-second MP3 clips, while lyria-3.5-pro-preview creates songs lasting a couple of minutes. Both output 44.1 kHz stereo audio, and the Pro model can also return WAV files. Image-based generation accepts a text prompt with up to 10 images.
Available across Google’s creative tools
Lyria 3.5 is available to all users globally through Gemini on the web and mobile. Google has also brought the model to Flow Music, Google AI Studio, Google Vids, and the Gemini API.
One prompt at a time
The current API workflow is single-turn, so it does not support iterative, multi-turn editing of an existing generated clip. Every track created with Lyria carries an imperceptible SynthID watermark.
Sources and checked date: Google Blog · Google DeepMind · Gemini API documentation · September 5, 2026
Leave a comment