Lyria 3.5
Audio modelCard summarizedGoogle's flagship music generation model
● Re-checked 12 Oct 2026 · 34 facts · 0 correctionsSource: Lyria 3.5 Model Card
Where to call it
- Gemini API
lyria-3.5
Good for
- High-quality music across a wide range of genres and styles
- Full-length songs with vocals and lyrics
Overview
Google · 03Lyria, front
Lyria 3.5
Google's flagship music generation model
Typical price
$0.08 / song
Full song
- Max length
- —
- Languages
- —
- Modes
- text-to-music, image-to-music (up to 10 images)
- Released
- 3 Sep 2026
Lyria 3.5 is a music generation system that makes audio and lyrics from a text prompt, using latent diffusion on temporal audio latents. Google says it is its most capable music model to date. The card reports no numeric scores, only that it improves significantly on Lyria 2 in audio fidelity and, with lyrics, in prompt adherence. Model information; Evaluation
lyria-3.5- Typical price
- $0.08 / song
- Weights
- Closed
- Released
- 3 Sep 2026
- text-to-music
- image-to-music (up to 10 images)
- vocals with timed lyrics
- custom lyrics with section tags
- instrumental