
A hamster, a cat, a chick, all from one company: a "pay-per-message chat" sock-puppet network on the US free Dating chart
What it is A US-only dating app: browse people nearby, play a 30-second mini-game of "catching" people popping out of hamster holes, chats cost coins, and swapping contact info requires a request.
Why it took off Not the product itself: the same Chinese team uses three developer identities to push a new mascot shell roughly weekly, rotates paid acquisition to burn US traffic, and recoups costs through pay-per-message chat. The US Play dating free chart has a low bar, and a few hundred installs a day gets it into the top 15.
EHow you could approach it Don't copy it. If you're building a legitimate dating social app, you could borrow a mascot and a 30-second mini-game as an icebreaker, and use the US dating chart's low bar to climb, but you first need real users in a narrow niche.
Est. monthly revenue (Google Play) at least $10KWe estimated this from its rank on country top-grossing charts and real install counts, calibrated against real revenue; typical error is 2–3x. Most apps that make $10K+/month get underestimated, hence "at least". Read it as a tier, not an exact number.
- Data (updated daily) · first launch 2026-10-26 · 11-20 installs 13,660
Android · Total installs worldwide Data 2026-11-04 → 2026-11-20 (17 days, 17 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 2026-11-05 → 2026-11-20 (16 days, 16 observations)
Luka is a dating app with a hamster mascot and a 30-second "whack-a-mole" style pick game, launched Oct 26. It reached 5,687 installs in two weeks and climbed to #13 on the US Google Play dating free chart. But what made it run wasn't the hamster: the same team put five re-skinned shells up within three weeks, only advertised in the US, and charged coins per message. This playbook was named in September by The Verge and Anthropic, and the previous batch of similar apps has already been taken down. This piece explains how it gets its numbers, where the money comes from, and why we don't recommend following it.
"This app and another are identical, one is a cat, the other is a bird"
In Oct 29, a user gave the dating app Waffle one star on Google Play: "This app is identical to another one. This one is a cat, that one is a bird. Same fake women, same fake explicit photos, reporting does nothing and no bans. Messages come at you like bots, one after another, and your coins (whatever it calls them) run out fast."
The "that one is a bird" he mentions is Rocco, published two weeks earlier by the same developer account, Silver State, with a chick mascot wearing aviator goggles. A few days later, Oct 26, the same account published a third: Luka, with the mascot swapped for a big-eyed round hamster.
What Luka is
The store page describes a "people nearby" dating app: browse nearby profiles, post updates, say hi. The selling point is Quick Pick: "Imagine whacking moles, except you're catching people you might actually want to meet" — people pop out of hamster holes, tap someone within 30 seconds to "catch" them, and afterward you decide whether to say hi. Hamster Luka is "half mascot, half wingman" (Play store page, 11-09).
What the store page doesn't say, but the privacy policy does, is more telling (lukaluv.com, effective 10-21):
- The currency is called Sparks, with gifts, a Plus Pass membership, and a paid subscription to see "who viewed me";
- Voice and video calls; exchanging contact info goes through a "request" flow;
- It logs "all user input and system prompts related to platform automation and intelligent processing systems";
- It uses AppsFlyer for ad attribution; "Luka is only for US users and is not promoted outside the US."
In-app purchases on Play run $3.99 to $189.99 each. The package name is lukapet and the subtitle is "Small friend, Big Happiness," which suggests it was first built as a pet app and later turned into a dating app.
How much it grew: 5,687 installs took it into the top 15 of US dating
| Date | Total installs | Score | US dating free chart | US dating top grossing chart |
|---|---|---|---|---|
| 11-04 | 9 | – | 96 | – |
| 11-05 | 331 | 9 | 36 | – |
| 11-07 | 331 | 31 | 17 | 61 |
| 11-08 | 2,009 | 41 | 18 | 55 |
| 11-09 | 2,962 | 58 | 10 | 44 |
| 11-10 | 5,687 | 71 | 13 | 41 |
Source: Google Play installs and chart rankings. Play install counts update only every few days; per third-party AppstoreSpy estimates, it adds roughly 300 to 840 installs a day. Rating 3.97 stars, with a steady 10 to 17 more reviews each day.
Two numbers worth noting. First, the bar for the US Play dating free chart is low: five or six hundred installs a day can land you in the top 15; BLK, at #12 on the same chart, grew 8,000 in a week, and Tinder, at #1, grew 785k in a week. Once you're in the top 20, you get browsing traffic from the dating category. Second, it also shows up at #36 to #72 on dating top grossing charts in a dozen-plus countries including Nigeria, India, Saudi Arabia, and Vietnam, but those rankings move in lockstep with the US ones at a fixed offset, and the privacy policy states US-only. Treat those small-country ranks as not real revenue (inference).
The real story is the five shells
Luka alone shows nothing. Put the same team's apps side by side:
| App | Developer account | Launched | Installs (11-10) | US dating free | US dating top grossing | Mascot |
|---|---|---|---|---|---|---|
| Rocco | Silver State | 10-13 | 44,978 | 5 | 15–21 | Aviator chick |
| Waffle | Silver State | 10-20 | 24,466 | 3 | 57 → 13 | Cat |
| Synca | IronRain | 10-22 | 0 | – | – | None |
| Blush | DOMETECHONE | 10-23 | 3,568 | 10 | 98 → 51 | None |
| Luka | Silver State | 10-26 | 5,687 | 13 | 76 → 41 | Hamster |
Source: Google Play, 11-10.
Three developer names, one team. What ties them together: the privacy policy domains for the five apps (lukaluv.com, roccovo.com, inwaffle.com, blushiay.com, syncaok.com) sit under the same Cloudflare account; they were registered at a Hong Kong registrar commonly used via Alibaba Cloud International, with Luka, Rocco, and Waffle's domains registered on the same day (10-14); all use Feishu for email; the feedback addresses follow the same pattern, lukaandfeedback@, waffleandfeedback@, Blushandfeedback@, syncaandfeedback@; and the contact Gmail on Rocco's and Luka's Play listings is the same. Even the store copy is cut from one mold: Waffle writes "Good connections often start with small things, a sunset on the way home..." and Synca writes "The best connections usually start with a small thing, a photo from a weekend walk..." (domain registration and DNS records, checked 11-10).
So the accurate description is: one Chinese team (inference: Feishu plus a Hong Kong registrar) uses at least three developer identities to push a new shell roughly every week. Luka is the fifth, and not the biggest. AppGazers' model estimates Luka's monthly Play in-app revenue at $10k to $46k, Rocco at $28k to $133k, and the Silver State accounts combined at $60k to $310k (estimates, wide range).
Each shell gets a different mascot, a small gameplay hook, a package name, and a domain. This serves two purposes: first, it lets Play treat them as different apps, slipping past "duplicate content" review (inference); second, it gives ad creatives a hook, since "catching people from hamster holes" looks more like a legitimate app than a photo of a pretty woman (inference, I haven't seen the creatives). The hamster is not why it took off.
Where users come from: paid acquisition, one shell after another burning through budget.
Off-platform buzz is zero: on X, TikTok and Reddit, searches for "Rocco / Waffle / Luka dating app" have been empty for the past 30 days. No press, no creators. The only hits are the Play pages and a few ranking sites (11-10).
So where's the traffic coming from? Most likely paid user acquisition (inferred; the ads themselves weren't captured). The evidence is indirect: the privacy policy names AppsFlyer, the standard attribution tool for paid-acquisition apps; the app only serves the US, which matches targeted ads; and the ramp is a relay: Rocco was pulling 3,000–4,000 installs a day from Oct 30, Waffle picked up in Nov 1, and Luka only started in Nov 5. That doesn't look like organic growth. It looks like budget being rotated across shells.
Ratings are the other helper. By the time Rocco had 6,670 installs, it had 342 ratings, about 5%; a typical app sits at 1% to 2%. Many of the five-star reviews are one or two words like "great", "cool", "awesome" (Google Play reviews). It looks like a rating prompt fired at the peak of user excitement (inferred).
Where the money comes from: mid-chat, the coins run out
The reviews spell out the pricing:
- "200 coins for $3.99, which gets you about 30 messages." (Rocco)
- "500 Sparks cost about $10, enough for a short chat. There's no unlimited-chat plan." (Waffle, 10-25)
- "It gives you the option to share your Snapchat or Instagram, but the other person never receives it. You pay to share and nothing happens." (Rocco, 10-29)
- "A girl actually calls you first on video. No other dating app does that." (Waffle, five stars, 10-24)
The playbook: you arrive and get swamped with likes, profile visits, messages and calls. Mid-chat the coins run out, and if you want to keep going you have to pay. If you try to leave the app to swap contact details, you get blocked. Nearly every bad review says the same thing: "all AI", "pure AI. Once you set up your profile, women from everywhere bombard you... the messages start sounding identical" (Rocco, 11-02), and "the women are paid to chat and video with you" (Waffle, 11-02).
To be clear: we haven't tested the apps ourselves, and reviews aren't proof. But a dating app two weeks after launch, with a few tens of thousands of installs, getting users "bombarded" and women calling first is something a real user base usually can't produce that fast.
This playbook has shown up before, and the last batch has already been pulled
In 2026 September, Anthropic published a case in its threat intelligence report, GTG-15001: a Chinese app studio used Claude to build 20 dating apps targeting only US users. About 75% of the recommendation feed was AI personas, and 25% were real people paid per task to do video calls and follow back on social accounts so users would believe the profiles were real. Chats and matches were metered against quotas and refilled with coins. To evade store review, the apps carried a UI that only switched on during review, and 20 variants deliberately renamed classes to dodge platform similarity checks. In two weeks, 4,700 AI personas and at least 25k people exchanged about 2.36M messages.
A follow-up by The Verge in September 16 named Dora, Doni, Romi, Luma, Jovia, Nalo and Kira: all cute short names, four letters or so. In the screenshots, Kira's coins were also called Sparks. These apps were pulled from the App Store and Google Play between late August and early September, but Kira was still live as of Sep 16.
What matches the earlier batch: a Chinese team, US-only, coins called Sparks, metered chat, video calls, paid profile visits, a request flow to exchange contact details, and a privacy policy that specifically mentions "system prompts". What doesn't match: the server domains and package names in the report don't match any of Luka's. So the honest claim is "same playbook, back a month after the last batch was pulled", not "same operators".
Who it's competing with
- Other shells from the same team: Rocco and Waffle launched one to two weeks earlier with bigger budgets, and reached #15–21 and #13 in US Top Grossing for Dating.
- The established pay-per-message chat matrix: Red Panda App's Only Spark and others, Evermatch, iHappy. They all use the store search pattern "Dating and Chat - [name]", each with 5M to 85M installs, and they've held #8 to #32 on the US Top Free Dating chart for years. This lane is already crowded.
- Real-person dating: Tinder (529M installs; its "Top Picks" is the prototype for Quick Pick) and Badoo (410M installs, a veteran of fast-swipe discovery).
- Mascot companionship: Finch (16.7M installs) shows that a little animal that keeps you company has emotional pull, but it isn't a dating app.
No serious competitor at scale in the store combines dating, a pet mascot and a 30-second mini-game. But that isn't evidence of an opening, because Luka itself didn't get traction from that combination.
If it were you
Don't copy it. Luka's traction comes from three things: a "other side" that makes new users feel surrounded from the first session even without real users, high-priced metered coins, and a release cadence of one shell a week. The first one is the part Anthropic and The Verge classified as fraud, and it's why the last batch was pulled. Remove it and what's left is a dating app that has to build a real user base from zero. Cold start is the problem it never solved for you.
- Who it fits: nobody should copy it as-is. If you're already building a legitimate dating or social product, three things are worth taking: a mascot plus a 30-second mini-game for icebreakers and store screenshots; the low bar of the US Play dating Top Free chart, where five or six hundred installs a day gets you into the top 15; and a way to run several brands off one backend and test which hook converts best each week.
- Who it doesn't fit: anyone who wants to clone metered paid chat, anyone who plans to fill cold start with AI personas, and anyone who only wants the US market with no real-user source.
- How to build it (for legitimate dating products only): first gather real users in one narrow place, a city, a circle, an interest. Then put a 30-second mechanic like Quick Pick into the daily feed to raise open frequency. Don't expect it to drive acquisition.
- How to acquire users: the US dating category has cheap chart positions. A small budget can push you into the top 20 of the Top Free chart and pick up category browse traffic. But that only amplifies: it works only if there are real people on the other side.
- Deadline: none. This isn't a window you need to rush.
- What would show this read is wrong: ① someone actually installs Luka or Rocco and confirms the chat partners are mostly real users. Then it's a legitimate dating app that grew on a mascot and mini-game, and worth reassessing. ② By 2027 February, Rocco, Waffle and Luka are still on Play and still in the US Top Grossing Dating chart's top 50. That would mean the platform isn't enforcing, and the team's unit economics are real. It still wouldn't change the conclusion to avoid the fake-people part. ③ A new app appears that doesn't rely on a matrix or fake people, only on the mascot mechanic, and holds a top 20 spot on the US Top Free Dating chart. That would mean the hamster mechanic has value on its own, and it's worth following.
Data and sources
- Google Play store pages, install counts, ratings, charts and reviews for Luka, Rocco, Waffle, Blush and Synca, 10-24 to 11-10. Daily new-install estimates come from AppstoreSpy.
- Privacy policy: lukaluv.com/privacy-policy (effective 10-21).
- Registration and DNS records for the five domains (registrar, registration date, nameservers, mail servers), checked 11-10.
- Revenue estimates: AppGazers developer page (Silver State). Model-based, not figures published by the company.
- Off-platform buzz: X, TikTok, web search, 11-10.
- Anthropic, "Detecting and countering misuse of AI: September 2026", GTG-15001 section; The Verge, "The sexy AI-powered dating app scams are here" (2026-09-16).
- Wherever we write "inferred" or "estimated", it's inference or estimation. We have not tested the apps ourselves.
Research notes: competitor table, acquisition evidence, full research report
- Competitors
Name Where Launched Scale How it differs from this one Rocco / Waffle (sibling shells on the same account) Android 2026-10-13 / 2026-10-20 45k / 24k installs; US Top Free Dating #5 / #3, Top Grossing #15–21 / #13 Same backend with a different mascot (pilot chick / cat), launched earlier with bigger budgets Blush: Meet Someone Real / Synca Android 2026-10-23 / 2026-10-22 3,568 / 0 installs; Blush at #10 on US Top Free Dating Same Cloudflare account and same feedback-email format, different developer identity, no mascot GTG-15001 network (Dora, Doni, Romi, Luma, Jovia, Kira, Nalo and others) Other 2025–2026,2026-08 ~09 pulled 20 apps; within two weeks, 4,700 AI personas sent 25M messages to 2.36k users (Anthropic) Same "US men + coin chat + AI / paid-fake" model, publicly named and pulled. Luka has no direct infrastructure link to it. Dating and Chat matrix (Red Panda App's Only Spark and others, Evermatch, iHappy) Android 2019–2026 5M–85M installs each, resident on US Top Free Dating at #8–32 Veteran pay-per-message chat matrix, all using "Dating and Chat - [name]" for ASO, no mascot or mini-game Tinder (Top Picks and others) Big-company features 2013 Play 529M installs, +785k a week Real-user network + subscriptions. Its daily picks are the prototype for Quick Pick. Badoo(Encounters) Android 2010 Play 410M installs Pioneer of gamified fast-swipe discovery, real users Finch: Self-Care Pet Android 2021 Play 16.7M installs Source of the emotional pull of pet companionship, but not a dating app - Where users come from
- Inferred Paid acquisition (US only): the privacy policy names AppsFlyer attribution and serves only the US. There's no organic off-platform buzz at all. 5 shells took turns driving the ramp (Rocco from 10-30 at 3–5K a day, Waffle from 11-01, Luka from 11-05). Our Google, Meta and TikTok ad libraries show no matching advertiser.
- Verified Play US Dating category chart: at roughly 500–800 installs a day it reached #10–13 on the US DATING Top Free chart (the same chart BLK gained +8K and Tinder +785k over one week), picking up category browse traffic.
- Verified Matrix shells (swapped mascot / package / domain / developer identity): lukaluv, roccovo and inwaffle were all registered at Dominet (HK) on the same day (10-14), with 5 domains sharing one Cloudflare nameserver pair and Feishu / Lark mail, feedback emails in the same <name>andfeedback@ format, and copy from the same template.
- Inferred Prompted ratings: by the time Rocco had 6,670 installs it already had 342 ratings (about 5%), with many one- or two-word five-star reviews. Luka consistently adds +10–17 ratings a day.
Why Luka (com.lukapet.app) took off
Bottom line: Nothing about Luka's product explains its surge. It's the third shell in a Chinese team's "dating skin farm." The same developer account, Silver State, shipped Rocco (10-13), Waffle (10-20), and Luka (10-26) within three weeks, all on the same Cloudflare account and the same Feishu business mailbox. Two other developer accounts, behind Blush and Synca, run on the same setup. Each shell swaps in a mascot (a pilot chick, a cat, or a hamster) and a small gimmick, targets only American male users, and monetizes with "pay per message in coins, no swapping contact info." Growth comes from paid acquisition plus the low bar of the Play US store's dating Top Free chart, not from word of mouth. The playbook closely mirrors the GTG-15001 "fake AI-person dating app network" that The Verge and Anthropic's threat report made public, and 2026-09 is at high risk of removal.
---
1. What it is
- Play name Luka, developer Silver State (the Play page lists Silver State Apps Inc, [email protected], 9805 Double R Blvd, Reno, NV, from the search-result summary of the Play page). Category DATING, first released 2026-10-26, version 1.1.0 (updated 10-30).
- Store copy: people nearby, a feed, and "Quick Pick", a 30-second whack-a-mole mini-game where people pop out of a hamster hole and tapping one "catches" them, after which you decide whom to say hi to. The mascot is Luka the hamster ("part mascot, part wingman"). Source: Play page.
- The privacy policy (lukaluv.com/privacy-policy, effective 2026-10-21) reveals more than the store page does:
- Virtual goods: Sparks (coins), gifts, Plus Pass membership, a "visitor-access subscription" (paying to see who visited you), Love Ring, Bond progress, and contact-exchange requests (swapping contact info goes through an approval flow);
- Voice / video calls, AR pets, shaking the phone to play with the pet; "a tally of days you and your companion have shared". The package name lukapet and the subtitle "Small friend, Big Happiness" show it was built as a pet app first;
- "all user inputs, system prompts, and algorithmic interactions tied to the platform's automated and intelligent processing systems" are logged. The AI insights feature generates titles on the fly via Azure OpenAI;
- Attribution runs through AppsFlyer. "Luka is run for users in the United States. We do not market, distribute, or tailor the service for people outside the U.S."
- In-app purchases $3.99–$189.99 each, with an ad flag (googleplay library, app_snap).
2. The data: how far it got
Data comes from the googleplay library (app_snap / app_day / chart_rank, 2026-11-04 to 11-10).
| Date | Total installs | Score | Average | US Dating Top Free | US Dating Top Grossing |
|---|---|---|---|---|---|
| 11-04 | 9 | – | – | 96 | – |
| 11-05 | 331 | 9 | 4.22 | 36 | – |
| 11-06 | 331 | 21 | 4.00 | 23 | 76 |
| 11-07 | 331 | 31 | 3.58 | 17 | 61 |
| 11-08 | 2,009 | 41 | 3.93 | 18 | 55 |
| 11-09 | 2,962 | 58 | 3.98 | 10 | 44 |
| 11-10 | 5,687 | 71 | 3.97 | 13 | 41 |
- Play installs update only every few days (331 stalled for three days in a row), so AppstoreSpy's daily estimates are more reliable: 11-05 about 318/day, 11-06~07 about 839/day, 11-08~09 about 476/day (app_day_ext, AppstoreSpy estimates). Ratings grow steadily by +10~17 per day.
- 11-10 star distribution [1★10, 2★3, 3★4, 4★10, 5★41] (app_snap). The library has no Luka review sample yet.
- The "14-country Top Grossing" claim is most likely an illusion: Luka's dating-category grossing ranks in ng / in / sa / vn / my / ph … move in lockstep with the US store, rising and falling together each day with a fixed offset (e.g. 11-10: us 41, ng 36, in 61, sa 61, vn 63 …). The same goes for Rocco and Waffle under the same account. The privacy policy also says it only serves the US. So the real money comes from the US, and the small-country dating grossing charts are the same aggregated chart (chart_rank).
- Third-party estimates: AppGazers' model values Luka at $9.7K–$46.4K (Play in-app revenue for one estimation period, P10–P90), Rocco at $27.8K–$133.3K, and Silver State's five apps combined at $62.5K–$309.4K/month (these are ranges added together, not company revenue). Source: AppGazers developer page.
Sibling apps from the same template (where it really shows)
| App | Developer account | Launched | Installs (11-10) | US dating free chart | US dating top grossing chart | Mascot / selling point |
|---|---|---|---|---|---|---|
| Rocco(com.roccoko.app) | Silver State | 10-13 | 44,978 | 5 | 21 → 15 → 20 (stuck at 15–21 since 10-29) | A pilot chick in goggles |
| Waffle — the cat has them all(com.prism.waffle) | Silver State | 10-20 | 24,466 | 3 | 57(10-29)→ 13(11-10) | A cat, "break the ice with posts" |
| Luka(com.lukapet.app) | Silver State | 10-26 | 5,687 | 13 | 76 → 41 | A hamster + Quick Pick whack-a-mole |
| Blush: Meet Someone Real(com.dometech.blush) | DOMETECHONE | 10-23 | 3,568 | 10 | 98 → 51 | "calmer place", no mascot |
| Synca - Meeting Your People(com.magicic.syncamore) | IronRain | 10-22 | 0 (never gained traction) | – | – | "share everyday moments" |
Data sources: googleplay library app / app_profile / chart_rank. Rocco's early curve comes from AppstoreSpy estimates: 10-23 already had 2,013 installs, and from 10-30 onward it averaged about 3,700–4,800 per day.
Evidence tying them together (my own DNS / RDAP lookups):
- The 5 privacy-policy domains (lukaluv.com, roccovo.com, inwaffle.com, blushiay.com, syncaok.com) use the same pair of Cloudflare nameservers (hunts / carol.ns.cloudflare.com). Cloudflare assigns that pair per account, so this strongly suggests a single Cloudflare account.
- All registered through Dominet (HK) Limited (a Hong Kong registrar often used via Alibaba Cloud's international site). lukaluv / roccovo / inwaffle were registered on the same day (2026-10-14), syncaok on 10-21, and blushiay on 10-23.
- Email all runs through Feishu / Lark (MX records are mx*.feishu.cn, while Blush and Synca use mx*.larksuite.com). The developer's main domain silverstateapp.com (registered at GoDaddy on 2026-03-05) also uses Feishu mailboxes.
- Feedback addresses follow the same pattern:
[email protected],[email protected],[email protected],[email protected]. Rocco's and Luka's store contact emails are the same Gmail address (silverdata15688@). - The store copy uses the same sentence structure. Waffle writes "Good connections often start with the little things. A sunset on the way home…", Synca writes "The best connections usually begin with something small. A photo from a weekend walk…", and the one-line descriptions all start with "Meet nearby singles, …".
Conclusion: this is one Chinese team (Feishu + Alibaba Cloud registrar) running at least 3 developer identities (US Nevada Inc, DOMETECHONE, IronRain), shipping a new shell every week. Luka is one of them, and not the biggest.
3. Why users come and why they pay
Where the users come from: paid acquisition + store chart placement (inferred, the ads themselves were not directly captured)
- The privacy policy names AppsFlyer for attribution, which is standard for paid-acquisition apps. Serving only the US also fits targeted ad buys.
- Off-store, there is no organic buzz to speak of. Jina / Brave only return Play pages, APKPure, and chart sites like Sensor Tower / Appfigures / Similarweb / AppBrain. x-radar over the last 30 days returned 0 results each for "Rocco / Waffle / Luka dating app" and "Silver State Apps". A TikTok search (TikHub) for "rocco / waffle / luka dating app" returned 0 results. There are no media, creator, or Reddit posts.
- Our ad libraries caught nothing: the Google Ads library lists Rocco and Waffle (added after they first crossed into the grossing chart at 11-05), but a reverse lookup on inwaffle.com found no advertiser (status = low, 0 advertisers). Nothing matched in the Meta or TikTok libraries. The TikTok library only collects ads targeted at Europe, and this app only advertises in the US, so "no match" proves nothing there. Google App Campaigns in-Play ads are hard to reverse-look-up by domain, so this point remains unconfirmed.
- The traction shape is one shell after another taking over: Rocco 10-30~11-05 at 3–5K per day, Waffle picked up from 11-01, and Luka only took off at 11-05. This looks more like rotating shells against a budget to test them than organic growth.
- Store chart placement is the amplifier: the bar for Play's US dating Top Free chart is low. Luka reached #10–#13 with 500–800 installs a day. Meanwhile BLK, at #12 on the same chart, gained 8K in a week, and Tinder, at #1, gained 785k. Once you crack the top 20, you get browse traffic from the dating category.
- The ratings look engineered. Rocco had 342 ratings (about 5%) by the time it hit 6,670 installs, and many of the reviews are one- or two-word five-stars ("great", "cool", "awesome") (googleplay library, review). It looks like the system rating prompt fires at moments of user good mood.
Why people pay: per-message chat billing
- Rocco's reviews spell out the mechanism: "pay 3.99 for 200 coin which lasts maybe 30 messages"; "limited messages and not being able to let people have your contact info"; "they monitor ur private messages so u cant exchange numbers"; "charged for another weeks worth of subscription". In other words: flood users with likes and visitors to hook them, charge coins mid-conversation, and block contact swaps so revenue stays inside the app.
- Nearly all the negative reviews are about bots and fake people: "All AI generated", "Bots that play with your feelings", "the women out there get paid to talk to guys", "100% fake… asking for a selfie verification seems to insult the bot" (Rocco, 11-05~07). Waffle has "another ai slop filled app" and "You need to pay to chat!".
- 同类商业模式的公开先例:Anthropic 2026-09 的威胁报告 GTG-15001: Deceptive dating app network 写的是:一个中国 App 工作室做了 20 多个约会 App,只针对美国用户;推荐流里约 75% 是 Claude 驱动的 AI 假人、25% 是按条拿钱的真人兼职;聊天按配额扣、用金币续;为躲商店审核,只在审核时激活一套 UI 控制器,20 多个变体故意改类名、逃避相似度检测。The Verge 2026-09-16 的报道点了 Dora / Doni / Romi / Luma / Jovia / Kira / Nalo 等名字,都是 4 个字母左右的可爱短名;截图里 Kira 的货币就叫「Sparks」;这些 App 在 2026-08~09 陆续被 Play 和 App Store 下架。
- Traits Luka shares with this network: a Chinese team, US-only, a coin called Sparks, metered chat, video calls, paying to see visitors, contact exchange requiring approval, and a privacy policy that mentions system prompts. But I have not installed the app to test it, and there is no code-level or infrastructure evidence linking Silver State to GTG-15001 (that report's indicators are the heyhru / archat.us domains and the packages com.qiga.vio and com.cavalier.nalo, none of which match here). So the honest claim is "same playbook, timed right after that batch was taken down", not "same people".
4. Competitors
| Competitors | Where | Launched | Scale | How Luka differs |
|---|---|---|---|---|
| Rocco / Waffle (sibling shells under the same account) | Android | 2026-10-13 / 10-20 | 45k / 24k installs, US dating Top Grossing #15–#21 / #13 | Same backend with a different mascot; launched one to two weeks before Luka, with a bigger budget |
| Blush / Synca (other same-template shells under two other accounts) | Android | 2026-10-23 / 10-22 | 3,568 / 0 installs | Same Cloudflare account, no mascot |
| GTG-15001 network (Dora, Doni, Romi, Luma, Jovia, Kira, Nalo…) | Android + iOS | 2025–2026 | 20 apps in total; within two weeks 4,700 AI personas, 25k users, 2.36M messages (Anthropic) | pulled 2026-08–09 after being publicly named; same "AI / part-time fake person + coin chat" model |
| the "Dating and Chat - X" matrix (Red Panda App's Only Spark / Lovector / Pheromance, Evermatch's Evermatch / Maybe You, iHappy) | Android | 2019–2026 | each with 5M–85M installs, likewise taking 20–30 spots on the US Play dating Top Free chart | an older pay-per-message chat matrix; no mascots, all using "Dating and Chat - Name" for ASO |
| Tinder (Top Picks, Swipe Night, etc.) | Big-company features | 2013 | 529M installs, +785k a week | a real-user network on subscription; "Daily Picks" is the prototype of Quick Pick |
| Badoo (Encounters fast swiping) | Android | 2010 | 410M installs | the veteran of gamified swiping |
| Finch: Self-Care Pet | Android | 2021 | 16.7M installs | the source of the "pet company" mascot emotional value, but not dating |
Bottom line: "dating + pet mascot + 30-second mini-game" has no sizable serious competitor in the stores; but the core that drives volume (pay-per-message chat + matrix shells + paid acquisition) is a crowded lane with several older players.
5. What's the differentiation, and is it why it broke out?
- The surface differentiation: each shell gets one mascot and one mini-game (Rocco's pilot chick, Waffle's cat and post-to-break-the-ice, Luka's hamster whack-a-mole Quick Pick), with copy pitching the "dating doesn't need a rulebook" lightness.
- This layer of differentiation is not why it ran. It does two real jobs:
- Passing review and avoiding association: Play has policies against duplicate content and spam matrices. Give each shell a different IP, a different set of copy, a different mini-game, a different package name (lukapet, roccoko, prism.waffle) and a new domain, and it can be listed as a "different app." This is the same idea as the Anthropic report's "20 variants with renamed classes to dodge similarity detection."
- Giving ad creatives a hook: "catch the person popping out of the hamster hole" looks more like a legit app than a photo of a pretty woman, so the paid creatives and store conversion may both do better (unverified, I did not capture the creatives).
- What actually made it run is operations: a release cadence of one shell a week, budget that can rotate to burn US traffic, and a high-ticket coin pricing ($189.99 tier) that earns back acquisition cost, plus the low bar for the Play US dating Top Free chart brings browsing traffic.
6. How many people have seen it
- Luka itself: almost nobody knows about it. No media, X, TikTok, or Reddit discussion, and revenue is not public; only Sensor Tower / Appfigures / Similarweb chart pages and AppGazers' modeled estimates (semi-public).
- But this kind of playbook has already been written up publicly: The Verge (2026-09-16) and the Anthropic threat report (2026-09) spelled out the network of "US men + coin chat + AI fake people" dating apps, and Google and Apple have the metrics. For Luka this is not a sign of "crowded with people" but a sign of a batch takedown that could come at any time.
- Copycats: none seen externally yet; Blush / Synca are shells from the same operator, not copycats.
7. What others can borrow
- A mascot plus a 30-second mini-game can make a store page and ad creatives look different in a homogenized category; a serious dating / social product could borrow it for "low-pressure icebreakers."
- The Play US dating Top Free chart has a low bar (roughly 500–1,000 installs a day gets you into the top 15), so a new dating app can catch category browsing traffic with a small amount of paid acquisition.
- One backend with many shells, one new shell a week, rotating budget to test brands, is a cheap way to test which IP / selling point converts.
- What not to borrow: using AI or part-time fake people to pose as real users and charging per message is the model Anthropic and The Verge have characterized as fraud, and it carries takedown and class-action risk.
8. Still unclear
- Whether the chat partners in Luka are AI / part-time fake people: not verified by installing the app; Rocco's many reviews say so, but reviews are not evidence.
- Acquisition channels and creatives: inferred to be Meta / Google App Campaigns, but none of our ad libraries matched.
- Whether Silver State / DOMETECHONE / IronRain and GTG-15001 are the same group: only the gameplay and timing line up; the infrastructure metrics do not.
- Revenue: AppGazers is only a modeled estimate; whether payments go through Play in-app purchase or a web checkout like GTG-15001's is not verified (the Play page does list in-app purchases of $3.99–$189.99).
- How much money the #41 spot on the US dating Top Grossing chart is worth and how long it lasts; the shell's lifespan (whether it gets pulled within two or three months like that network).
Sources
- googleplay library (agent-im machine DuckDB): app, app_snap, app_day, app_day_ext (AppstoreSpy estimates), chart_rank, review, app_profile, 2026-10-29 to 11-10.
- googleads / metaads / tiktokads libraries: watch, advertiser, app_link (no matching advertiser found).
- Play pages: play.google.com/store/apps/details?id=com.lukapet.app, play.google.com/store/apps/developer?id=Silver+State
- Privacy policy: lukaluv.com/privacy-policy
- AppGazers:appgazers.com/reports/developers/android-Silver+State
- Anthropic threat report (2026-09): www.anthropic.com/threat-intelligence-report-september-2026#…
- The Verge(2026-09-16):www.theverge.com/ai-artificial-intelligence/995348/ai-dating…
- DNS / RDAP: dig NS / MX, rdap.verisign.com (looked up 2026-11-10)
- x-radar, TikHub, Jina, Brave search (2026-11-10; none had off-store discussion)