Google DeepMind Ships Lyria 3.5 Inside Flow Music
*Google’s newest music model claims better musicality, lyrics, vocals, and user control, but supplies no benchmarks or examples.*
Google DeepMind and Google both announced the release of Lyria 3.5 on 29 July 2026. The model is available only inside the Google Flow Music product. The companies state that it improves musicality, lyric generation, vocal quality, and creative control compared with earlier versions.
What the announcements say
The two posts use nearly identical wording. They describe Lyria 3.5 as “our newest music generation model” that “delivers significant advancements across musicality, lyrics, and vocal quality, empowering you to craft richer tracks.” No technical specifications, training data details, or performance numbers appear in either source.
Limited public information
The releases contain no side-by-side comparisons, no audio samples, and no statements from external musicians or researchers. They also omit any mention of safety filters, licensing terms for generated output, or plans for wider API access.
Why it matters
Engineers and producers who already use Flow Music will see the new model appear in their existing workflow. Everyone else receives only a marketing claim until independent tests or third-party access become available. The absence of concrete metrics leaves the scale of the claimed improvements unknown.
---
Sources:
No comments yet