Google DeepMind’s music generation model Lyria 3.5 has been made globally available to Gemini users and developers following Google Flow Music. Compared with Lyria 3, the model constructs complex melodies more effectively and offers precise adherence to prompts, high pronunciation accuracy, and realistic vocals. Users can select genre, vocal, and instrumental options, make use of ready-made templates, and adjust track length. The model is available through the Gemini app, Gemini API, the web, mobile, Google AI Studio, and Google Vids.
Why it matters
The significance of the development is that the model brings both end users and developers together around the same music generation technology without being limited to a single application. From a developer perspective, the API and Google AI Studio diversify the channels through which the model can be tested in different use cases. On the user side, the available settings and templates allow people with varying levels of technical knowledge to work more flexibly in music generation. However, it is not yet clear to what extent the announced features will prove useful in practice or under which conditions the difference compared with Lyria 3 will become more apparent.
Background
Google is not a new name in the FikirPilot archive: we have published 29 news stories mentioning the name in the last 90 days; the most recent is dated September 7, 2026.