Each song costs 0.14 cents: a free, unlimited AI songwriting app built by a Pole on an open-source model
What it is An AI songwriting app that takes a one-line description and writes lyrics, adds vocals, and produces a full song. It has a web version that needs no sign-up, is free with ads, and offers a $5.99/month subscription to remove ads.
Why it's taking off It runs on an open-source model, so each song costs a fraction of a cent. Watching one ad covers the cost, which lets it be truly free while most competitors sit behind subscription walls. Users mostly arrive through multilingual long-tail search pages on the web version, then get pushed into the app.
CHow you could follow it Don't build another AI songwriting app. Move this structure to a generation category that just got open-source commercial-use models and hasn't been filled by subscription walls or paid acquisition yet: a no-sign-up web version + multilingual long-tail SEO pages + unlimited use with ads + a "hosted version for people without a GPU" pitch in model communities.
- Data (updated daily) · first launch 2026-09-11 · 11-07 installs 24,429
Android · Total installs worldwide Data 2026-10-29 → 2026-11-07 (10 days, 8 observations). Google's detail page only refreshes install counts every other day, so a flat day doesn't mean nobody installed.Android · Ratings count (cumulative) Data 2026-10-29 → 2026-11-07 (10 days, 8 observations)
Boppy is an AI songwriting app that generates a full song from one line. It runs on an open-source model, so each song costs a fraction of a cent. That lets it be free, require no sign-up, and keep generating after one ad. In a category full of subscription walls and paid acquisition, it pulls people into the app through multilingual long-tail search pages on its web version, reaching 24K installs in two months. Growth is healthy but small, and the moat is shallow. What's worth learning is its cost structure and acquisition path, not the category itself.
One post, one blog
In Mar 2026 14, under the official demo page for ACE-Step 1.5, the open-source songwriting model on Hugging Face, a user named deAstro posted: "For those having trouble running it on ZeroGPU or locally — Boppy is a free hosted version." For people who couldn't run it locally or get a free GPU slot, Boppy was a free hosted option.
Half a year later, in Sep 11, Boppy's Android version launched on Google Play. The same day, inference API provider deAPI published a post on its official blog, "AI App Development Made Easy: 4 Reasons to Build on deAPI," using Boppy as a launch case and doing the math: "$5 of credit, which is roughly 3,490 two-minute tracks." $5 generated about 3,490 two-minute songs, or roughly $0.0014 per song.
By Nov 6, the counter on Boppy's website showed 722,807 songs generated in total. At that unit cost, the model cost for over 700,000 songs is roughly $1,000 (estimate).
What it is
Open the app or the website and describe a song in your own words: "breakup ballad," "birthday song for a friend," "lo-fi for studying," or hold the mic button and say it. It writes lyrics, adds vocals and backing, and produces a full song in about a minute. You can edit the lyrics before generating. Each song comes with cover art, scrolling lyrics, and an MP3 download, plus an automatically created public link you can drop into WhatsApp or a group chat. Lyrics support 10 languages, and the interface supports 15.
做的人是 Konrad Sierzputowski,波兰的连续创业者,3D 打印机厂商 Omni3D 的联合创始人,现在在做一家视觉 AI 公司(LinkedIn)。Boppy 的前端是用 AI 生成 App 的工具 Rork 搭的,同一个开发者账号下还有说唱版的 Rapper - AI Rap Maker,以及几个别的小 App:一套后端,换个题材再发一个。
Where the money comes from
Two revenue legs, mostly ads:
- Free version: a short ad plays before each song. The store page and official site both say "no daily song limit for normal use" (abuse prevention has separate limits). One user wrote on Google Play: "freaking awesome, i love how it is free. all you have to do is watch an add then done! 5 stars" (9-28).
- Premium: a monthly subscription at $5.99/month in the US store, with no free trial. It removes ads, allows 200 songs per month, songs up to 5 minutes long, and voice cloning (Google Play store page and update notes, 11-06).
This only works because of those 0.14 cents. A rewarded video ad usually earns several times that amount (inferred), so watching one ad covers the cost of one song. Competitors mostly do the opposite: they hit you with a subscription wall on the first song, or give a few credits and then charge.
Revenue isn't visible from outside. It hasn't charted in any country's Top Grossing.
How big it has gotten
| Date | Google Play cumulative installs | Source |
|---|---|---|
| 09-30 | About 2,100 | Third-party estimate |
| 10-13 | About 9,300 | Third-party estimate |
| 10-24 | About 15,200 | Third-party estimate |
| 10-31 | 18,793 | Google Play |
| 11-06 | 24,429 | Google Play; past 7 days +8,364 |
In October it added a few hundred installs a day, and after late October it reached roughly a thousand a day. The curve has no single day spiking into the tens of thousands, which doesn't look like concentrated paid buying (inferred).
1,464 ratings, averaging 4.87 stars. Ratings equal about 6% of installs, which is high for Google Play. Most of the 2-star and 3-star reviews are 0, likely because the rating prompt popped up right after users heard their first song (inferred). Negative reviews are specific: "I need European Portuguese, but it always comes out with a Brazilian accent" (9-26), "too many ads" (9-27).
On the charts it has only appeared in Poland's Music & Audio Free chart, ranked between #146 and #198 (10-29 to 11-06). The developer is Polish, and the store page is localized into Polish. The iOS version launched in June and has 114 ratings in the US store.
Where users come from
The main channel is the web version and search. boppy.me lets you generate songs directly with no sign-up. Its sitemap has 2,854 URLs across 20 languages, including English, Polish, Spanish, Portuguese, German, Korean, Chinese, Russian, and Vietnamese. Each language has about 130 landing pages, and each page targets one long-tail search term: "Free AI Music Generator No Sign Up," "No Watermark," "With Free MP3 Download," "Text to Song." There are also blog posts like "Best AI Music Generator for YouTube" and "Royalty-free music for YouTube." The Google Play download button on the landing pages carries a dedicated source tag (utm_source=boppy.me, utm_campaign=web_app_landing_hero), so routing web users into the app was designed on purpose. About 200 public share pages for user songs are also indexed by search engines.
The open-source model community. That's the post at the top: people who want to use ACE-Step but don't have a GPU are naturally its users.
The API provider's case-study blog. deAPI's post brought it a bit of exposure and also publicly explained its approach to everyone.
Occasional organic recommendations. A small Hindi-language YouTube channel posted "Make songs in any singer's voice for free with Boppy, no copyright issues," which got 229 views, with a download link in the description. A few AI tool directories list it, with very little traffic.
Searching TikTok for "boppy ai" returns only unrelated accounts with the same name, like penguins and babies. No one is discussing it on X or Reddit, and there's no media coverage. The ad library shows no ads from it, but not finding ads doesn't mean it isn't running them.
Who it's competing with
This is a very crowded category.
| Product | Google Play scale (11-06) | How Boppy differs |
|---|---|---|
| Suno | 50.4M installs, +7k in the last 750 days | Own model, best quality; requires sign-up, charges once credits run out |
| Donna AI Song & Music Maker | 13.6M, +176k/7 days | Subscription wall, paid acquisition |
| AI Hit Music Maker | 5.55M, +430k/7 days | Paid-acquisition type, spread across 27 countries |
| Beatron, AI Music & Song Generator, etc. | Around 5M, about +200k per week | Subscription wall wrapper |
| Melozi | 630k, +148k/7 days | Launched around the same time as Boppy, late August. Growing 25× faster, most likely through paid acquisition |
| Web tools like acestep.io | Not verified | Same underlying model as Boppy, both on ACE-Step |
Boppy gets about a thousand installs a day, while a same-period paid-acquisition competitor gets twenty thousand. Its difference is in business model and acquisition, not product capability: anyone can plug in the model, and deAPI's blog is effectively a tutorial for copying it.
If it were you
C. You can learn a very cheap playbook here, but don't build another AI songwriting app. The top of this category is Suno, and below it a row of paid-acquisition machines each pull tens to hundreds of thousands of installs a week. Boppy's "truly free" approach is right, but it has only 24k installs after two months, its ad revenue per song is in cents, and a follower copying it wouldn't do much better.
Who it suits: indie developers who want to build tools on "open-source models + cheap inference," especially those whose first language isn't English and who can write solid localized pages; anyone who already has a generation product and wants a way to acquire users without ad spend.
Who it doesn't suit: anyone who wants to head-on compete for money in AI songwriting (you can't match Suno's quality or outspend paid-acquisition apps); anyone relying on ad revenue to support themselves (at this scale of installs, ad revenue is likely a few hundred dollars a month, estimate).
How to do it: move the structure to a generation category not yet filled by subscription walls or paid acquisition. Find a model that just came out, works well enough, and is open-source with commercial use allowed. Use an inference API to get the cost per use below a cent, and build it as "no sign-up, unlimited with ads, optional monthly subscription." Build the front end in a week with tools like Rork or Lovable, then launch several apps on one backend across different themes. Ask for a rating right when users get their finished output.
How to get users: Ship a web version first, where people can get results without signing up. Build landing pages in 10+ languages targeting pain-point long-tail keywords like "free", "no signup", "no watermark", "commercial use OK", "for YouTube", with one download button per page tagged with its source. Turn user creations into public share pages so search engines index them. Then show up on this open model's Hugging Face demo page and GitHub discussions as a "hosted version, no GPU needed". Find inference API providers and pitch to get featured in their case-study blogs.
Deadline: the sooner the better. The first few months after an open-source model comes out are the most valuable. Once a new model is released, the hosted-version and SEO-page spots likely get filled within a few months (inferred).
A miscalculation: ① Boppy passed 200k installs by late January in year 2027 without any paid acquisition. That means SEO plus a genuinely free product can compete head-on with paid user acquisition, and the AI song-writing category is still worth getting into. ② It hit the top grossing chart in at least one country. That means the $5.99/month subscription converts better than expected, so this model is more than just ad revenue. ③ On the flip side, if a Google search update knocks programmatic multilingual pages like boppy.me out of results, this acquisition playbook has to be re-evaluated.
Research notes: competitor table, acquisition evidence, full research report
- Competitors
Name Where Launched Scale How it differs from this one Suno Android Play 2024-12 (web 2023) 50.4M installs, +7k in the last 750 days Suno uses a proprietary closed-source model, runs on credits, and requires sign-up. Boppy uses an open-source model, needs no sign-up, and lets you generate unlimited songs by watching ads. Donna AI Song & Music Maker Android 2024-04 13.6M installs, +176k/7 days Subscription wall + paid acquisition. Boppy is ads + optional subscription, relying on organic traffic. AI Hit Music Maker(Goldlab Pro) Android 2023-12 5.55M installs, +430k/7 days A paid acquisition machine covering 27 stores. No sign of paid acquisition for Boppy. Beatron - AI Music Maker Android 2025-11 5.26M installs, +178k/7 days A subscription-wall wrapper. Boppy is free with unlimited generations. Melozi: AI Song, Music Maker Android 2026-08-27 630k installs, +148k/7 days Launched around the same time as Boppy but growing 25x faster, most likely through paid acquisition. Boppy relies on SEO: slower, but cheaper. Boomy / DeepAI Music / acestep.io Web 2019–2026 Not verified A web tool. acestep.io is also a wrapper around ACE-Step. Boppy adds an app and a multilingual SEO page matrix. ACE-Step 1.5 (open-source model) Open source Early 2026 Open-source model with GitHub/HF/ComfyUI You need your own GPU and have to write your own prompts. Boppy is managed for you: one sentence gets you a song. - Where users come from
- Verified Web version + multilingual programmatic SEO: boppy.me's sitemap has 2,854 URLs across 20 languages, each with about 130 long-tail landing pages (no sign up / no watermark / for YouTube, etc.). The Play badge carries utm_source=boppy.me.
- Verified Traffic from the open-source model community: 2026-03-14 HF user deAstro recommended Boppy as a "free hosted version" in the discussion section of the official ACE-Step Space.
- Verified API provider case study: deAPI's official blog 2026-09-11 wrote up Boppy as a launch case study.
- Verified Organic recommendation from a small YouTube creator: the Hindi channel Gana Likho posted a video with 229 views and a Play link.
- Inferred Polish local store search / category charts: it has only charted between #146 and #198 on the Polish Music & Audio free chart. The developer is Polish, and the store listing is localized into Polish.
- Inferred Paid acquisition: the install curve shows no spikes, and the ad library has no coverage, so this can't be confirmed.
Boppy: AI Music Generator (app.boppy.android): how it pulled this off
Channel: Google Play | Developer: Konrad Sierzputowski (Poland, individual developer account) | Live on Play for 2026-09-11 (iOS version 2026-06-25) | Research date 2026-11-06
Bottom line
Boppy is an AI song-making app that wraps an open-source model. The front end is generated with Rork, and the back end calls ACE-Step 1.5 XL Turbo, an open-source model hosted by deAPI.ai. A two-minute song costs about $0.0014 to generate. That's a fraction of what Suno costs, which is why it can offer "free, no sign-up, no daily limit, and keep generating as long as you watch ads." Most of its users come from the web version, boppy.me: about 2,850 pages and programmatic SEO across 20 languages, with download badges on the pages that send people to the app. It's still small (24k installs, around 1,000 per day), and nobody is talking about it.
1. Data (our googleplay / appstore libraries, plus AppstoreSpy estimates)
| Date | Total installs | Notes |
|---|---|---|
| 2026-09-30 | 2,096 | AppstoreSpy estimate (app_day_ext) |
| 2026-10-13 | 9,343 | Same as above |
| 2026-10-24 | 15,215 | Same as above |
| 2026-10-29 | 16,065 | Our app_day table |
| 2026-10-31 | 18,793 | Our library |
| 2026-11-03 | 22,814 | Our library |
| 2026-11-05/06 | 24,429 | Our library; last 7 days +8,364 |
- In October, it averaged about 450–600 new installs a day. After late October, that rose to about 800–1,300 a day (10-30 had +2,728 in one day). The curve has no spike of 10,000+ in a single day, which suggests no concentrated ad buying or chart pushing.
- There are 1,464 ratings, roughly 6% of installs, which is high for Play. The average is 4.87 stars, with a distribution of [31, 0, 0, 62, 1370] (1–5 stars). The 2- and 3-star ratings are both 0, which likely means the app prompted users for a rating at the moment they heard their song for the first time (inferred, not verified).
- Charts: it has only charted on the Polish Music & Audio free chart, ranking between #146 and #198 (chart_rank, 10-29 to 11-06). It has never appeared on a grossing chart. The iOS version only shows up in the category charts of small storefronts like Poland, Ukraine, and Austria, around #100–200. The US store has 114 ratings (iTunes lookup). The official site claims 4.7 stars and 306 ratings on iOS.
- Monetization: the Play listing says "Contains ads," and in-app purchases are $5.99 per item. The update notes say "Premium: ad-free, 200 songs every month." The official FAQ says Premium includes: ad-free use, a fixed monthly quota, songs up to 5 minutes long, and Voice Clone.
- A counter on the official site shows 722,807 songs generated to date (boppy.me Android landing page, 2026-11-06 read).
2. Competitors
| Competitors | Platform | Launched | Scale (googleplay library, 11-06) | How Boppy differs |
|---|---|---|---|---|
| Suno | Android / iOS / web | Web 2023, Play 2024-12 | 50.4M installs, +7k in the last 750 days, charted in 30 stores | Suno uses a proprietary closed-source model with the best quality, but its free tier is limited by credits and requires sign-up. Boppy needs no sign-up, and you can generate unlimited songs by watching ads. |
| Donna AI Song & Music Maker | Android | 2024-04 | 13.6M, +176k/7 days | A classic subscription-wall wrapper. Boppy uses an "ads + optional subscription" model. |
| AI Hit Music Maker(Goldlab) | Android | 2023-12 | 5.55M, +430k/7 days | A paid acquisition machine covering 27 stores. No sign of paid acquisition for Boppy. |
| AI Music & Song Generator(blekitek) | Android | 2024-06 | 5.41M, +200k/7 days | Same as above |
| Beatron - AI Music Maker | Android | 2025-11 | 5.26M, +178k/7 days | Same as above |
| Melozi: AI Song, Music Maker | Android | 2026-08-27 | 630k, +148k/7 days | Launched around the same time as Boppy but growing 25x faster, most likely through paid acquisition. |
| Rythmix / AI Song & Lyric Generator / OurSongs, etc. | Android | 2024–2026 | 500k to 6M | All subscription walls |
| Boomy, DeepAI Music, acestep.io, etc. | Web | 2019–2026 | Not verified | Web tools. acestep.io is also a wrapper around ACE-Step, with exactly the same underlying model as Boppy. |
| ACE-Step 1.5 itself (GitHub / HF / ComfyUI) | Open source | Early 2026 | Open-source model | You need your own GPU and have to write your own prompts. Boppy turns that into a single sentence. |
Bottom line: this category is already crowded. Suno sits at the top, and a long tail of subscription-wall wrappers has reached millions or even tens of millions of installs through paid acquisition. Boppy isn't a major player by scale. It's a small product with a fairly healthy start.
3. Where users came from (growth tactics)
- Web version + multilingual programmatic SEO (confirmed). boppy.me's web version generates songs with no sign-up. The sitemap contains 2,854 URLs in total, split across 20 languages (en/pl/es/pt/de/ko/zh/ru/vi, and more), with about 130 pages per language. There are also about 200 public /s/ song-share pages. Landing page titles target long-tail search terms, such as "Free AI Music Generator No Sign Up," "No Watermark," "With Free MP3 Download," "Text to Song," and "Pop Song Generator." There are also blog posts like "Best AI Music Generator for YouTube" and "Royalty-free music for YouTube." The Google Play badge on the landing pages carries
utm_source=boppy.me&utm_medium=badge&utm_campaign=web_app_landing_hero, showing that the web-to-app funnel was deliberately designed. - Riding the open-source model community (confirmed). 2026-03-14, HF user deAstro posted in the discussion section of the official ACE-Step Space: "For those having trouble running it on ZeroGPU or locally — Boppy is a free hosted version…" (huggingface.co/spaces/ACE-Step/Ace-Step-v1.5/discussions/9). The account name includes "de," but whether it's connected to deAPI hasn't been verified.
- Case study from the API provider (confirmed). 2026-09-11 (the same day it launched on Play), deAPI's official blog published "AI App Development Made Easy: 4 Reasons to Build on deAPI," presenting Boppy as a launch case study. It even dissected the bundle id
app.rork..., showing it was built with Rork. - Listed in AI tool directories (confirmed): relvehq.com, aikc.vn, and the fusoelektronique forum all list it, but traffic is very small.
- Organic plug from a small Indian YouTube channel (confirmed, very small reach): the channel "Gana Likho" posted a Hindi video titled "Boppy ai app se kisi bhi singer ke aawaj mein gane banao free no copyright," published 4 weeks ago, with 229 views and a Play link in the description. The pitch is "free, no copyright."
- Store search and Polish roots (inferred): on Play, it has only charted in the Polish category charts. That fits the developer being Polish and the store listing being localized into Polish ("Generator Muzyki AI"). However, Play install counts are global cumulative totals, and the installs outside Poland most likely come from SEO and store search. This remains an inference.
- Channels with no evidence found: searching TikTok for "boppy ai" only turns up unrelated accounts with the same name; nothing from Boppy itself. X (x-radar corpus, last 90 days) returned no hits. Reddit has no discussion. No media coverage found. The ad libraries show nothing (our libraries only track products on our watchlist, so a missing result doesn't mean no ads were run).
4. Why it took off
- Cost structure makes "free and unlimited" work. deAPI's blog put it this way: "$5 of credit, which is roughly 3,490 two-minute tracks," which works out to about $0.0014 per song. Ad revenue from a single rewarded video is enough to cover the cost of one song, so the app can promise "no daily song limit" (per the official FAQ). Competitors are mostly subscription walls or credit systems that start asking for money after the first song.
- Zero friction: the web version needs no sign-up, no card, and no watermark, and its terms explicitly allow commercial use (Terms §8). These points speak directly to the search terms users type most, like no sign up, no watermark, royalty free, and for YouTube.
- Retention and ratings: 4.87 stars, with ratings at 6% of installs. Users get a complete, shareable song (with a share link and MP3 download), which drives high satisfaction and sets up self-spreading. The share pages /s/ can also be indexed by search engines, which in turn feed back into SEO traffic.
- The developer is an experienced serial founder with resources. LinkedIn shows Konrad Sierzputowski has 20 years of startup experience. He co-founded Omni3D (a 3D printer maker) and is now building Vised (a visual AI agent). He uses Rork to ship apps in bulk. The same account has other apps: Rapper - AI Rap Maker (com.rork.rapper), Disaster Generator, NoBand, and Reactor Bike.
5. Differentiation, and whether that's why it broke out
The difference lies in the business model and acquisition, not the product. It uses open-source models, monetizes through ads, promises free unlimited use, and adds a no-signup web version plus multilingual SEO. In a category full of "subscription wall + paid acquisition", it's one of the few that's actually free, and that's a real reason it's taking off. Web SEO brings organic traffic, "free and unlimited" earns high ratings, and high ratings help the store convert. But this differentiation moat is shallow. Anyone can plug in the same base models (acestep.io and others are already doing the same), and deAPI even publishes a blog post teaching people to copy the setup. Its current scale (about 1,000 new installs a day) is one to two orders of magnitude behind paid-acquisition competitors (Melozi at 20k a day). It's "small but healthy", not yet a breakout.
6. Visibility and copycats
- Visibility: few people know about it yet. No public revenue, no press coverage, no discussion on X or Reddit. The only "endorsement" is the API vendor's marketing blog.
- Copycats: so far the lookalikes are just similar products, with no clear evidence of anyone deliberately copying it. acestep.io and other ACE-Step wrapper sites are on the same path; the developer used the same stack to build Rapper - AI Rap Maker; and the deAPI blog is effectively urging others to replicate the model.
7. Takeaways
- Open-source models plus cheap inference APIs push the per-generation cost below one cent, then "ads + free unlimited use" attack the "first paywall" of subscription-wall competitors.
- A no-signup web version delivers results directly, with multilingual long-tail SEO landing pages and a UTM badge that funnels users to the app. User-generated share pages feed back into SEO.
- Drives traffic from open-source model communities (HF Space discussions, GitHub) by pitching a hosted version that needs no GPU.
- Partners with API vendors to get into their case-study blogs.
- Triggers a rating prompt at the moment the user receives their finished output, to push ratings up.
- Uses tools like Rork to mass-produce apps: same backend, different theme (rap, disaster films), another launch.
8. Uncertain
- Install country distribution is unknown. The Poland chart only shows Poland is one piece; where the global volume comes from is inference (SEO + store search).
- Whether it runs paid acquisition: ad libraries don't cover it. From the curve shape, it looks unlike paid acquisition.
- The relationship with deAPI: the blog reads like a third-party view, but the HF account promoting it, deAstro, has not been verified, so a paid partnership can't be ruled out.
- Premium conversion and ad revenue are completely unknown. It's also unclear whether $5.99 is the in-app purchase price or the subscription price.
- The star distribution is unusual, with 2 and 3 stars at 0. It may be how Google handles small samples, or it may be rating prompting; uncertain.
Sources
- This library: googleplay (app_profile / app_day / app_day_ext (AppstoreSpy estimates) / chart_rank / app_snap / app, pulled from 2026-11-06), appstore (app_profile)
- boppy.me/, boppy.me/free-ai-music-generator-app-android, boppy.me/sitemap.xml
- deapi.ai/blog/ai-app-development-made-easy-4-reasons-to-buil…(2026-09-11)
- deapi.ai/models/ace-step-15-xl-turbo
- huggingface.co/spaces/ACE-Step/Ace-Step-v1.5/discussions/9
- www.youtube.com/watch?v=IPPNfbt-1sU
- www.linkedin.com/in/konrad-sierzputowski, www.crunchbase.com/person/konrad-sierzputowski
- play.google.com/store/apps/developer?id=Konrad+Sierzputowski, apps.apple.com/us/developer/konrad-sierzputowski/id187158872…
- iTunes lookup id=6778658290
- TikHub TikTok search and x-radar search (both returned no relevant hits)