Lyria 3.5 is Google DeepMind's music-generation model for songwriters and producers who want to turn a text prompt into a full song with vocals, instrumentation, and production in a single pass. Released on July 29, 2026 inside the Flow Music platform (part of Google Labs), it is the third major Lyria release in five months.
Core Features
- Full-song generation: synthesizes vocals, instruments, and effects together from one prompt.
- Better musicality: more natural harmonic progressions and continuity across verses and choruses.
- Coherent lyrics: follows prompts more closely and holds song structure across sections.
- Expressive vocals: noticeably more natural pronunciation and emotional inflection.
- Creative controls: adjust tempo, key, instrumentation, and vocal style, and set the length of the generated audio.
- SynthID watermarking on every output for traceable, AI-labeled audio.
Use Cases / Best For
- Independent artists drafting demos and scratch tracks quickly.
- Content creators making royalty-free background music for video.
- Producers prototyping hooks and arrangements before studio work.
Pricing
Flow Music offers a free tier with limited credits and a premium tier that unlocks higher-resolution audio and longer editing tools; Google has not published fixed per-song pricing. It sits alongside Suno and Stability AI's music models in the generative-audio field.
Our Take
Best for songwriters who want rapid, controllable prototypes inside Google's ecosystem; the trade-off is that Flow Music's exact commercial licensing terms are still evolving, so verify rights before releasing a track. Explore more in our AI Audio tools directory.
