Google's web and iOS AI music studio, now running Lyria 3.5 — DeepMind's upgraded music model with more natural vocals, better lyrics, and direct tempo and duration control.
Last verified · 2026-07-31 · by Moe Ameen
Google Flow Music is Google's dedicated AI music studio — a browser app at flowmusic.app and an iOS app where you generate full songs from a text prompt, an image, or your own lyrics. It is powered by Lyria, Google DeepMind's music generation model, and on July 29, 2026 Google rolled Lyria 3.5 into it as the default engine. The same Lyria line also shows up elsewhere in Google's stack — Gemini, YouTube Shorts' Dream Track, Google Vids, and Google AI Studio for developers — but Flow Music is the purpose-built interface for making a complete track.
Lyria 3.5 is a quality upgrade rather than a new product. Google frames the gains in four areas: musicality (richer, more complex melodic structures that hold up across a full track instead of looping mechanically after the first section), lyrics (better prompt adherence and structural awareness), vocals (more realistic, emotionally nuanced singing with cleaner pronunciation), and creative control (you can now steer tempo and duration more directly). It is the third Lyria milestone in 2026 — Lyria 3 arrived in February and Lyria 3 Pro followed in late March with tracks up to roughly three minutes — and for existing Flow Music users it landed as a free upgrade with no new tier and no waitlist.
Access is tied to a Google account. There is a free plan with daily credit top-ups and a couple of concurrent generations, plus paid Flow Music plans (reported around $8/month Starter, $24/month Plus, and $64/month Member, scaling how many songs you can make); Google AI Plus, Pro, and Ultra subscriptions include the equivalent Flow Music tiers. Confirm current numbers on Google's live pricing page, since they move. Every track Lyria produces is watermarked with SynthID so it can be identified as AI-generated or AI-edited.
The honest framing: Flow Music makes an excellent song — and, in its Spaces, a matching music video and cover art — then stops there. It writes no social captions, builds no captioned per-platform short-form, holds no brand voice across a week of posts, and publishes to no social platform — the track lands in your Google account, and everything after that is on you.
Lyria 3.5's headline win is that the vocals and lyrics are finally good enough that a Flow Music track can front a post instead of just sitting under B-roll — but a song in your Google account reaches no one on its own. Music travels on short-form video, and the jump from "I generated a track" to "this is a captioned, publish-ready video people actually watch and share" is precisely the work Kompozy does. Export your Lyria 3.5 song and bring it into Kompozy as the audio bed for the formats built for music: Naturalistic Video and Listicle Video lay your title and body cards over a portrait clip, Clipped Shorts cut a longer piece into several standalone vertical moments, and Persona Frames composite a face-locked avatar into a brand-exact HyperFrames template — with captions burned in, the frame reframed to 9:16, 1:1, and 16:9, and the whole set held to one voice by your Persona Brief.
Then one track becomes a launch. Kompozy fans a single music idea into 25–35 outputs — the short itself, a Carousel that teases the release, Quote Graphics pulled from the lyrics, native Text Posts, a Blog Article about the drop, and an Email Newsletter to your list — and schedules and publishes them across nine platforms plus blog and email from one queue with Autopilot and a per-post review pass. Because Kompozy is source-agnostic, the pipeline doesn't care that the audio is AI-generated and SynthID-watermarked: if a monetized post needs a licensed track instead, you swap the soundtrack without rebuilding a single video. Flow Music scores the song; Kompozy turns it into a presence.
Google Flow Music is Google's AI music studio (web at flowmusic.app and an iOS app) where you generate full songs from a prompt, an image, or your own lyrics. Lyria 3.5 is the DeepMind music model that powers it — rolled in on July 29, 2026 as a free upgrade with better vocals, lyrics, musicality, and tempo and duration control.
Google cites four areas: more natural, complex melodic structures (musicality), higher-quality lyrics with better prompt adherence and structure, more realistic and emotionally nuanced vocals with cleaner pronunciation, and easier control over a track's tempo and duration. It is a quality upgrade to the existing model, not a new product.
There is a free plan with daily credit top-ups and limited concurrent generations. Paid Flow Music plans (reported around $8, $24, and $64 per month) raise how many songs you can make, and Google AI Plus, Pro, and Ultra subscriptions include the equivalent tiers. Lyria 3.5 itself is a free upgrade for existing users. Confirm pricing on Google's live page.
Yes. Every track Lyria generates is embedded with SynthID, Google's imperceptible watermark, so the audio can be identified as AI-generated or AI-edited. That matters for platforms that badge or de-monetize AI music, so plan your disclosure and swap in a licensed track where a monetized post needs one.
Flow Music makes the song — and, in Spaces, a music video for it — but not the captioned, per-platform short-form or the supporting posts. Export the track and bring it into Kompozy, where it becomes the bed for social-ready video — Naturalistic Video, Listicle Video, Clipped Shorts, or a Persona Frames avatar composite — with captions and per-platform reframing, then Kompozy fans the idea into a carousel, quote graphics, posts, a blog, and a newsletter and publishes across nine platforms.