190k installs in a month, a fraction of Suno: how Google Flow Music's Android version took off

Google Flow Music is an AI music studio from Google Labs. It hit Google Play in Oct 5, got 192k installs in a month, and climbed to #46 on the US Music Top Free chart. It grew on Google's own models, brand and subscriptions, plus a round of creator promotion, none of which indie developers have. Measured against its category, its volume is actually modest. This piece covers what it is, where the volume came from, and which two moves a small team can take.

An official account with only 8,752 followers, four videos each with over a million views

In Sep 2026 22, the TikTok account @googleflowmusic posted a video with singer Annalise Azadian, building her a custom vocal effect plugin called "ANNA•DOTE". That one got about 1.22M views. Three days later, on Sep 25, the same account posted three more in the identical format, all tagged #ad: Caleb Colossus made a plugin that "takes his voice up a level"; J☆URNEE built her own vocal effect plugin; Alina Smith recreated her usual vocal chain, which she normally says costs over a thousand dollars, as 3 free plugins. Those three got about 1.05M, 1.08M and 1.3M views.

At the time the account had only 8,752 followers. Its other videos perform far worse: the Jul 31 video announcing the Lyria 3.5 model got 1,862 views, and the Oct 6 video with Grammy-winning producer Khris Riddick-Tynes explaining new features got 672. The gap is three orders of magnitude, and the likes are low too (the 1.3M-view video has only 842 likes). It's a strong inference (high confidence) that these four were paid boosts. In Oct 30 it also ran a German collaboration video labeled "Anzeige" (ad).

Less than two weeks after these four videos finished running, the Android version launched.

What it is

An app for writing and editing songs with AI. You give it a prompt or a set of lyrics, it generates a full song, and then you revise it as if talking to a producer: "more bass", "sing it lower". You can also upload your own tracks to swap vocals or style, split stems, and use Google's video model to make a music video for the song.

Its most distinctive feature is called Spaces: describe the effect or instrument you want in plain language, and it "writes" it for you, then exports it as a VST3/AU plugin you can load into professional music production software. The four creator videos above were promoting exactly this.

Its origin is worth knowing. It started as Riffusion, was renamed ProducerAI, and in 2026 Feb 24 Google's official blog announced it was joining Google Labs; in Apr 20 it was renamed Google Flow Music (9to5, Google coverage). The model is Google DeepMind's Lyria, upgraded to Lyria 3.5 at the end of July.

Pricing: the core features are free, with a small daily credit top-up. Paid tiers: Starter $8/month (3,000 credits, roughly 600 songs), Plus $24, Member $64 (prices compiled by third-party review sites; annual plans get a discount). The key point: subscribers to Google AI Plus, Pro, or Ultra get the matching tier included. Pro users get 10,000 credits per month (Google One help page).

Numbers: flat, steady, not big

Google Play installs (scraped daily):

Up 7 installs over 44,517 days, averaging just over 6,000 a day. In the first 24 days it averaged about 5,000 a day, and it hasn't accelerated or dropped since. On the US Music & Audio free chart it climbed from #10-29 (70) to #11-05 (46). It has charted in 14 countries' category charts, all around #80 except the US. It hasn't charted on the overall or Top Grossing charts. Ratings are only about 0.27% of installs, with no sign of review farming.

Compared with its peers:

(Google Play,11-06)

About 45k a week, roughly 6% of Suno's and 2.5% of Google Flow's. For Google that's far from a hit, just a steady small slice of a category that's already been proven. The iOS version launched in May 18, and on the US store it still has only 59 ratings, with a best of #206 on the music free chart. So Android is where the real volume is.

Why now: the Android version that took five months

Laid out as a timeline: in February it joined Google, in April it was renamed, in May Google I/O announced a mobile app but only for iOS, with Android listed as "coming soon"; at the end of July Lyria 3.5 launched; in late September it ran a creator video campaign; in Oct 5 the Android version finally went live.

Google's official blog, 9to5Google, CNET, and MusicTech all covered this path, and Reddit threads like "Will Flow Music kill Suno?" and "Flow Music is what Udio was supposed to be" were discussed for several rounds. Android users who had waited five months had a ready-made crowd to install it on launch day (inference).

The Android slot was also open. Udio never shipped on Android, and at the end of October 2025 it also disabled downloads on all plans while it pivots into a licensed platform with Universal Music. Before this, there was basically no good non-Suno model to use on an Android phone.

Where users come from

Ranked by how much is confirmed:

1. Google's own exposure: the official blog, I/O, and an entry point on the Labs homepage (labs.google), with tech press following up. Confirmed.
2. Subscription bundling: people already paying for a Google AI subscription get several thousand to tens of thousands of credits on install, with no extra payment. Confirmed.
3. Free credits and "vote for credits": no card needed for core features. It has a mechanic called Turntable: listen to two samples, vote, and earn credits (Google Flow help page). Inviting friends also earns credits. Confirmed.
4. TikTok creator campaign: the four posts above, totaling about 4.6M views. The spend is certain, but how many Android installs those views drove can't be seen.
5. The "Google Flow" name: the sibling AI video tool Google Flow has 2,000 million+ installs on Play, and people searching "Google Flow" may have installed the music version by mistake. Inference only; I haven't verified whether the video app pushes users to it.
6. Organic word of mouth: lots of YouTube tutorials titled "How to use Google Flow Music (Free)" and "Suno vs Flow Music." Reddit's take is roughly "sound quality is good, but little control (Suno is similar)." On Hacker News people complain it can't remix and the tone drifts.

No store reviews have been pulled yet, so user retention and paid conversion aren't visible from outside.

Who it's competing with

AI songwriting is a crowded category. Suno leads, with annual revenue above $300M and over 2M paying users (Music Business Worldwide, 2026-10). Its pitch is "one sentence, one song" plus a community. Below it sits a large crop of wrapper apps (Mozart, Donna, VocalMe, MyTunes and the like) that use other vendors' models, set hard paywalls, and sell weekly subscriptions. On the US iOS store, a single one can collect thousands to a hundred thousand ratings. ElevenLabs has its own music product too. Google's Gemini app also generates short tracks with Lyria.

Flow Music differs from them in three ways: it runs on its own top-tier model rather than a wrapper; it leans toward a producer's toolkit (section edits, covers, stem splitting, plugins) rather than one-click songs; and it's bundled into Google subscriptions with cheaper credits than Suno. But the thing that actually drives volume is the Google resources mentioned above. These differences rank lower.

If it were you

D: this product can't be copied. Its moat is its own frontier model, the Google brand, subscription bundling, and Google's store. Any follower gets none of that. And AI songwriting is already saturated by Suno and a flood of wrapper apps, so even Google only pried loose a sliver of Suno's share.

Who it's for: people who want to learn the playbook, not people trying to build the same product. Two moves a small team can take:

• Turn user preferences into credits. "Listen to two, vote once, earn credits" lowers the barrier for free users and collects a pile of preference data along the way. Any per-generation product can do this.
• Make creator content "exclusive," not just "demo." Instead of creators demoing "generate a song," each one made a vocal plugin named after themselves. Creators want to post it, and the content doesn't all look the same. Tool products working with creators can borrow this approach.

Who it's not for: anyone thinking of building another general-purpose AI songwriting app. The model belongs to someone else, your rivals are Suno and Google, and you can't afford the free credits.

If you really want to build something nearby: don't fight head-on. Go after positions that big companies and leaders haven't covered yet, or covered late. Flow Music itself is an example: it came in when Android had no good second option. The method is to first map where the category's top products are missing by platform, language, or audience, then build that slice on a licensed music-model API (custom song gifts, children's songs in a specific language, and background music for short videos are all ready-made directions).

What would show this read is wrong: if Flow Music's weekly installs climb past 200k in the next month or two, or it breaks into the US overall chart, that would mean it's getting more than "Google users installing it out of habit," and its producer-toolkit route may actually work on the consumer side. In that case the AI songwriting category gets more crowded, not more open.