25 AI app builders compared, from idea to App Store: what each does, what it costs, and when to hire help

Making your own app has never been this easy to start. You type what you want into an AI app builder, and a few minutes later something runs on your phone. The App Store saw about 600,000 new apps in 2025 and about 560,000 in the first half of 2026 alone, while downloads grew only 2% (Sensor Tower figures via the New York Times, July 2026). Building got cheap. Getting an app through review, onto people's phones and found is still work.

We compared 25 builders that a person with no coding background can start from a prompt or a visual editor: their plans and free tiers, what they actually produce, how far each gets you toward the App Store and Google Play, what users say in reviews, and which apps we could find that really shipped. Then we walked the whole path from idea to store submission, twelve stages, to show which ones a builder covers, which you can do alone as a vibe coder, and where an hour of a freelancer is cheaper than another week of prompts.

Prices are the vendors' list prices in USD a month, read on 18 and 19 October 2026, billed monthly unless marked yearly. They change often, so check the vendor page before you pay. "Credits" are each vendor's own unit and are not comparable across tools. (3p) marks a figure we could only find from a third party.

The 25, in four groups

The tools sort into four groups, and the group tells you more than the brand:

Mobile-first prompt builders make real iPhone and Android apps from a prompt: Rork, a0.dev, Vibecode, Newly (was Natively), Anything (was Create), Dreamflow (from the FlutterFlow team) and Superapp. Most are a year or two old.

Web-first prompt builders make web apps: Replit, Lovable, Bolt, v0, Base44, Emergent, Blink and Google AI Studio's Build mode. A few reach the phone through Expo (a React Native toolchain); the rest need a wrapper, and a wrapped website runs into Apple's minimum functionality rule (more below).

Visual builders with AI added are the older drag-and-drop tools, most with store publishing built in: FlutterFlow, Bubble, Adalo, Thunkable, Draftbit, Glide, Softr, Bravo Studio and Appy Pie.

The developer baseline is Cursor, an AI code editor. It is the "do it yourself as a developer" row: it cuts the hours of someone who can code, it does not do the store steps for anyone.

Three builders that disappeared this year

Mocha shut down on 1 August 2026: sign-ups closed, subscriptions were cancelled and users were pointed to Anything (Seedtable). Google stopped new Firebase Studio workspaces on 22 June 2026 and will delete the rest on 22 March 2027 (Firebase). GitHub Spark took no new users after 4 August and access ended on 31 August (GitHub changelog). Three of the 2025 crop gone within a year is the best argument for one column in the tables below: whether you can take the code with you.

The big table: plans, free tiers, output and the store path

"Store path" is what the tool does toward the App Store and Google Play. "Code" is whether you can export the source and carry on without the tool; "?" means the vendor doesn't say. "cr" is credits.

Mobile-first prompt builders

Web-first prompt builders

Visual builders with AI added

Developer baseline

Three things stand out. First, store submission is almost never free: Rork's free tier has no build mode, Vibecode needs Plus, Blink needs its $200 Max plan, and FlutterFlow, Bubble, Adalo, Thunkable, Draftbit and Bravo all put publishing on a paid plan. Second, the popular web builders don't make phone apps: Lovable, v0 and Base44 make websites, and Bolt, Replit and Emergent reach the phone only through Expo, where the signed build is mostly yours to make. Third, only about half let you take the code, which matters when a builder changes its pricing or closes.

Higher plans are often steep: Rork Max starts at $200 and goes to $1,800, Replit Pro is $100, Emergent Pro $200. Sources for every row are at the end.

What users say

We read Trustpilot for every tool and a second source (G2, Capterra, Product Hunt or the vendor's forum) for most. Trustpilot leans toward people with a billing grievance; G2 and Capterra lean toward business users who were asked to review. Many "reviews" you'll find in search are written by a competing builder, so we used those only for what they quote. Scores and counts as of 19 October 2026.

The complaints that come up again and again

Paying for the tool's own mistakes. Credit burn on fix loops is the top complaint for every prompt builder with a Trustpilot page. On Replit: "The AI bot would fix one thing and break another to cost more and more" (18 September 2026) and "I paid $440 after repeated out-of-credit prompts while trying to get the project working" (13 September 2026) (Trustpilot). On Emergent: "I burned through 100 credits in about an hour and got zero working results out of it" (8 October 2026) (Trustpilot). Even Lovable, the best liked, hears it after its free credits shrank: "Before, you could do SO MUCH with 5 daily credits, now it just runs out before anything has been done" (17 October 2026) (Trustpilot).

Refunds and cancelling. Base44, Bolt, Vibecode, Blink, Adalo, Anything and Glide all have "no refund" or "charged after cancelling" reviews. Pay monthly, not yearly, until you know the tool, and cancel the day a project is done.

Lock-in. Rork reviewers lost access to projects after cancelling; Adalo and Thunkable apps stop working when the plan lapses; Adalo, Bravo, Base44 and Glide have no or weak code export.

The ceiling. Most reviews say the same thing in different words: these tools are great up to a first version and struggle once sign-in with roles, payments or custom logic arrive. That is where the freelancer line below comes in.

Old visual builders score far better on G2 than on Trustpilot (Glide 4.7 against 2.3, Bubble 4.4 against 2.9, FlutterFlow 4.5 against 2.5). Businesses that stay are happy; individuals who hit a paywall at publishing are not.

Who actually ships to the stores

This was the hardest part to research. Almost no vendor publishes a list of store apps you can check, and most "apps built with X" lists come from agencies selling X builds. What we could verify:

Strong evidence exists only for the older visual builders and for web apps. FlutterFlow's customer stories include the Ab.Money meditation app (4.8 on the App Store, 250,000 users, built by an agency) and the Postaj handyman marketplace (32,000+ downloads across both stores) (FlutterFlow). Adalo runs a #MadeInAdalo showcase with apps in both stores (Adalo). Lovable's success stories are web products, not store apps. Note who built the FlutterFlow examples: mostly agencies and developers, not first-time solo builders.

Some evidence: Replit documents a real path (an Expo project, then a Launch wizard that signs the build and uploads it to App Store Connect for TestFlight) (Replit docs), and RevenueCat has a full guide for adding store billing to Replit apps. Bubble's native mobile is newer and its older forum examples are mostly gone. Draftbit offers a managed publish on paid plans.

Thin evidence: the 2025-26 prompt-to-phone builders (Rork, a0.dev, Vibecode, Newly, Anything, Dreamflow, Superapp, Blink) all have a working publish button but almost no shipped apps we could check. That is not proof they don't work; it means you are an early user. The NYT story above names two vibe-coded apps without naming the tool: InfoDrizzle (1,300 downloads, about $500 profit after five weeks of work) and Stampa (20,000 downloads, about $3,000 profit after a week).

When Apple turns on the builder itself

Replit, Vibecode, Anything, Emergent and Rork all have iPhone apps you can build in. Apple's guideline 2.5.2 says an app may not "download, install, or execute code which introduces or changes features or functionality of the app, including other apps". In March 2026 Apple blocked updates for Replit and Vibecode, and it removed Anything twice (TechCrunch, 9to5Mac). This hits the phone app you build in, not the app you build with it, but it is one more way a builder can change under you.

The store rules that catch AI-built apps

Whatever the tool, your app goes through the same review. These are the rules that come up most for apps from builders, in Apple's and Google's own words (App Review Guidelines, Google Play testing requirements):

Apple 4.2, minimum functionality: "Your app should include features, content, and UI that elevate it beyond a repackaged website." This is the rule that catches wrapped web apps from Lovable, v0, Base44, Glide or Softr. Every guide gives the same advice: add something only a phone app does, such as offline data, the camera, push notifications or Face ID.

Apple 4.2.6: "Apps created from a commercialized template or app generation service will be rejected unless they are submitted directly by the provider of the app's content." In practice: submit under your own Apple developer account, never the builder's or a freelancer's.

Apple 4.3, spam: "Don't submit apps that are indistinguishable from what's already widely available." Generic AI-made habit trackers, to-do lists and quote apps hit this one.

Apple 3.1.1, in-app purchase: "If you want to unlock features or functionality within your app, (by way of example: subscriptions, in-game currencies, game levels, access to premium content, or unlocking a full version), you must use in-app purchase." A web builder's Stripe checkout inside a wrapped app does not pass. Google Play has the same rule for digital goods.

Apple 4.8, login services: an app that uses Google, Facebook or another social sign-in for the main account must also offer "an equivalent option": a sign-in that limits data to name and email, lets people keep their email private and doesn't track them for ads without consent. Sign in with Apple is the usual answer.

Apple 5.1.1: a privacy policy link in App Store Connect and inside the app, and "If your app supports account creation, you must also offer account deletion within the app."

Apple 2.3.3: "Screenshots should show the app in use, and not merely the title art, login page, or splash screen."

Google Play closed testing: personal developer accounts created after 13 November 2023 must run a closed test with at least 12 testers opted in for 14 days in a row before production is unlocked, then answer an "apply for production" form about how testers were found and what changed. Google reviews it in about a week and can ask for more testing. This hits every first-time solo Android app, whichever tool built it.

From idea to store: twelve stages, and who does each

Here is the whole path. "yes" means the tool does it inside the product, "part" means it helps but you finish it, "no" means it is outside the tool. Mobile, Web and Visual are the three builder groups above.

Cursor, the developer row, covers stages 3 to 6 and 9 in code. Every other stage is the same as for the builders: an AI code editor cuts a developer's hours, it does not remove the research, the policy pages, the testers, the screenshots or the store forms.

Stages 1 to 3: idea, journey, spec

No builder checks whether the idea is already in the store. Do it yourself, and it costs nothing: search the store for your idea, read the one-to-three-star reviews of the top five rivals, and check when they were last updated. It is also your best defence against Apple's 4.3 "already widely available" rejection.

Then write the user journey in plain words before the first prompt: who opens the app, the one job it does, and what happens in the first 60 seconds. Follow it with a one-page spec: the screens, the data each one shows, and who can see what. Rork, Vibecode, Lovable and Replit have planning modes that help, but vague prompts are where the credit-burning fix loops above start, so this page is the cheapest credit saver you have. A UX designer for two to five hours is worth it if the app has onboarding, payments or several kinds of users; an hour of a developer checking the data model saves rebuilds later.

Stages 4 and 5: mockups and building

This is what the builders are good at. Rork's design mode, Superapp's Figma import, v0 and Lovable for web UI, Bravo from Figma, FlutterFlow and Draftbit visually: generate the screens, then click through them on a real phone, not only in the browser preview. A vibe coder can build the happy path, simple lists and forms, and one or two kinds of data alone. A designer for the icon and brand look is a small spend ($10 to $200 for an icon on Fiverr).

Stage 6: the server side

Sign-in and a simple database work out of the box in most tools (Lovable Cloud, Bolt Cloud, Base44, Replit, Vibecode Cloud, Rork Cloud, Bubble, Adalo; FlutterFlow and Dreamflow through Firebase or Supabase). Store billing, push notifications and Sign in with Apple mostly are not. This is the most common hire: row-level security so users can't read each other's data, Apple and Google billing (usually through RevenueCat), push certificates and webhooks. Fiverr is full of "connect your Rork or Lovable app to Supabase" gigs for a reason.

Stage 7: the web site and the policy pages

Both stores ask for a privacy policy URL and a support URL, Apple wants in-app account deletion, and Google wants a web page where people can ask to delete their account. The mobile-first builders don't make these pages. AppBro does exactly this step: one form, and you get a home page, privacy policy, terms, support and delete-account pages on HTTPS, ready to paste into App Store Connect and Play Console. You only need a lawyer if the app handles health, children's or financial data.

Stages 8 and 9: pilot and testing on real phones

Rork, Superapp and Vibecode submit to TestFlight for you; Replit and Bolt get there through Expo; FlutterFlow, Adalo and Draftbit build and upload. On Android, start looking for your 12 testers early: they must stay opted in for 14 days in a row, and Google asks how you found them, so paid tester gigs are a risk. Test on your own phone plus one older, small-screen phone of the other platform, and try no network, a denied permission and a deleted account. If the app takes money, a QA tester for a few hours before the first release is cheap insurance.

Stages 10 and 11: screenshots, listing and submission

No builder makes good store screenshots; Rork's publish flow makes icons and listing text, per reviews. Take real screenshots of the app in use (Apple 2.3.3) and put them in a free frame template, or pay $5 to $200 on Fiverr once the app works. Then the forms: description, privacy labels, Google's Data safety form, age rating, and reviewer notes with a demo login. Fill them in under your own developer account. Fiverr "publish your app" gigs run about $100 to $200 for Android only and $500 to $800 for both stores (seller listings); avoid any that publish under the seller's account. Before you press submit, run the AppBro store check for the links and pages reviewers look for.

Stage 12: updates

Every builder keeps your app updated only while you pay, and some lock projects or code after cancelling. Keep the plan as long as the app is live, or pick a tool with code export (FlutterFlow, Dreamflow, Draftbit, Newly, Anything, Vibecode Pro, the Expo-based ones) so a developer can take over when you hit the ceiling.

The freelancer line: when to stop prompting and hire

You can be a total vibe coder for most of the path. Book a freelancer, even for one hour, when one of these is true:

1. Money moves in the app. In-app purchases, subscriptions, refunds. Apple and Google billing has to be right, and a Stripe checkout in a wrapped app gets rejected.

2. You hold other people's data. Sign-in with roles, row-level security, account deletion that really deletes.

3. You are in a fix loop. The same bug survives three prompts, or a day, or about a month's plan in credits. The reviewers above spent $440 and 100 credits in an hour this way; an hour of a freelancer usually costs less.

4. The store said no twice. A 4.2 or 4.3 rejection is usually about the product, not the build, and a second opinion is cheaper than a third rejection.

5. The tool hits its ceiling. The builder cannot do a feature at all. With code export a developer carries on from there; without it you rebuild.

Keep for yourself: the idea, the journey, the spec, the store research, the tester list, the story your screenshots tell, and the developer accounts. Those need judgment about your customer, not code.

What freelancers charge (October 2026)

These are marketplace listings and rate guides, not a survey, so treat them as ranges:

The honest summary: a few focused hours from a mid-level developer at $50 to $150 an hour cover most of the triggers above. Marketplaces add their own buyer fees (Fiverr 5.5%, Upwork 3% to 10%).

What year one costs, before any freelancer

That is roughly $370 to $850 in the first year for a solo builder in both stores, then $99 plus the plan every year after. One trap from the table: a free Supabase project pauses after a week without activity, and a backend that is asleep while Apple reviews your app gets it rejected.

On commission: Apple charges 15% on paid apps and in-app purchases for developers who made up to $1 million the year before and for those new to the App Store (Apple Small Business Program); Google charges 15% on the first $1 million a year and on subscriptions (Google Play service fees). Both default to 30% above that.

Which builder for which person

You want a real phone app and have never coded: start with a mobile-first builder or Adalo, use the monthly plan, and check that store publishing is on the plan you buy. Adalo has the best evidence of apps actually reaching both stores; the newer prompt builders are faster to a first version but have little shipped-app history, so prefer one with code export.

You want a web app, or a phone app later: Lovable is the best reviewed of the prompt builders. Plan for a wrapper and real native features if you go to the stores, or use Replit's Expo path.

You want to grow it into a business with a developer later: FlutterFlow, Dreamflow or Draftbit, because the exported Flutter or React Native code is something a developer can carry on with.

An internal tool for your team: Glide or Softr. No store needed, and these get the best support scores.

You can code a little: an AI code editor like Cursor with Expo gives you the most control and no lock-in, and the store stages are the same as everyone else's.

Whichever you pick, the twelve stages don't change. The tool does stages 4 and 5 well; the rest is research, pages, testers, screenshots and forms that you can mostly do yourself, plus a freelancer at the five points above.

Sources

Vendor pricing and docs (read 18-19 October 2026): Rork, a0.dev (plans via Techjockey), Vibecode, Newly, Anything, Dreamflow, Superapp, Replit, Lovable, Bolt, v0, Base44, Emergent, Blink, Google AI Studio, FlutterFlow, Bubble, Adalo, Thunkable, Draftbit, Glide, Softr (via Jet Admin), Bravo Studio, Appy Pie, Cursor.

Reviews (read 19 October 2026): Trustpilot pages for each tool (trustpilot.com/review/<domain>), G2 FlutterFlow, G2 Emergent, G2 Bravo Studio, Capterra Draftbit, Vercel Community on v0 credits.

Store rules and fees: Apple App Review Guidelines, Apple Developer Program enrollment, Apple Small Business Program, Google Play testing requirements, Google Play registration, Google Play service fees, Expo pricing, RevenueCat pricing, Supabase pricing.

Shipped apps and builder news: FlutterFlow Ab.Money, FlutterFlow production apps, Made in Adalo, Replit to the App Store, RevenueCat for Replit, Draftbit publishing, Bubble native app listing, NYT via The Star on vibe-coded apps, TechCrunch on Anything, 9to5Mac on Apple and vibe-coding apps, Mocha shutdown, Firebase Studio sunset, GitHub Spark deprecation.

Freelancer rates: Second Talent, goLance, Chop Dawg, Fiverr gig listings.