AI App Builders Reviewed: Lovable, Base44, Bolt, Replit and v0 Compared
Every one of these platforms will take a sentence and hand you a running application. That part is settled, and it works. What separates them is everything that happens afterwards: what the bill looks like in month three, whether you can take the code somewhere else, and whether the app is open to the internet by default.
They also get lumped into one category when they belong in three. Hosted prompt-to-app builders (Lovable, Base44, Bolt, Replit) generate and host the whole thing. UI-first generators like v0 hand you code and leave hosting to you. Editor and terminal agents such as Cursor and Claude Code sit inside a real repository and assume you can read the output. Security researchers draw the line in the same place, and for a reason: the editor tools require a human to review code before it ships, while the builders generate, execute and deploy with almost no oversight in between.
Lovable
Lovable is the money leader. It raised $400 million in August at a $13.3 billion valuation, exactly double the $6.6 billion it carried in December, after hitting $500 million in annualized revenue in June. The company says 60 million projects have been created on the platform, drawing 900 million monthly visits. It now runs its own in-house trained model alongside the usual frontier options, and signed a multiyear Google Cloud agreement in June representing a fivefold jump in usage.
Pricing is credit-based. Free, Pro at $25 a month, Business at $50, enterprise starting somewhere around $500, with metered cloud hosting layered on top and top-ups available when the monthly allotment runs dry. That structure is why revenue scales faster than seat count; the heaviest users pay disproportionately.
What you get is the best-looking output in the category, a visual editor, Figma import, GitHub sync, and a choice between Supabase and Lovable’s own managed backend. Where it stalls is complex business logic. If you already know your way around a full-stack codebase, Cursor will move faster and cost less.
Base44
Base44 is the shortest distance between a sentence and a live URL. Wix bought it in June 2025 for roughly $80 million in cash with up to $170 million in earn-outs running through 2029, six months after the product came out of stealth. It reports over two million users, was tracking toward roughly $150 million in annualized revenue by mid-2026, ran a Super Bowl ad in February, and shipped its own Base1 model in June.
The pricing is where people get caught. There are two separate credit pools. Message credits pay for building with the AI; integration credits pay for what your live app actually does, meaning emails, API calls, uploads, model calls. Neither pool rolls over, and you cannot buy more of either without upgrading your whole plan tier. Published plans run $20 to $200 monthly, or $16 to $160 annually.
Two structural facts matter more than the price. The mobile output is a web view rather than a native build, and you do not own the runtime your app runs on. Long-time users report effective costs running 15 to 30 percent higher per equivalent app since the acquisition, and getting data out is genuinely difficult. The review scores split in a way that tells its own story: 2.8 out of 5 on Trustpilot against 4.4 on Product Hunt. That gap is usually the sound of delighted first-week users and unhappy month-three users.
Bolt.new
StackBlitz built Bolt on WebContainers, which run a real Node environment inside the browser. No cloud VM spins up, so previews appear faster than anything else here. Bolt V2 added Bolt Cloud, bringing databases, authentication, file storage, edge functions and hosting into the platform and closing the deployment gap that earlier versions left open. A Microsoft Azure partnership in May 2026 added enterprise procurement and own-tenant deployment.
Free gives one million tokens a month capped at 300,000 daily. Pro starts at $25 for ten million tokens with no daily cap, and Teams is $30 per member. Unused tokens roll over for one additional month, which nothing else in this comparison offers.
Two things make Bolt the portability pick. Claude Code and OpenAI Codex were wired into the platform directly in September 2025, letting you switch coding agents inside the same project. And bolt.diy, the open-source fork maintained on GitHub, lets you bring your own API keys for any model and run the whole thing locally.
Replit
The most capable platform on this list, and the one with the sharpest edges. Agent 3 takes a plain-English prompt and produces a full-stack app with schema, authentication, frontend and deployment, typically in under an hour. It spawns sub-agents, writes and runs its own tests, and connects to more than 160 third-party services through OpenInt, which Replit acquired in October 2025.
The same platform deleted a founder’s production database and generated four-figure weekly bills for users. Core runs $25 a month ($20 billed annually) with $25 of usage credits; Pro is $100 with faster models and credit rollover. Pricing is effort-based, so a small fix costs under $0.25 while a complete app build runs $5 to $20. G2 and Capterra reviews average around 3.7 out of 5, and the complaints cluster tightly around money: credits burning faster than expected, bill shock on heavy weeks, and annual plans that issue credits monthly rather than pooling them.
What you get in return is control. A real shell, direct file editing, git, package management, secrets, multiple deployment targets, and native mobile tooling, all on one screen.
v0 by Vercel
v0 rebranded from v0.dev to v0.app in January 2026 and now claims more than six million developers. Free comes with $5 in monthly credits, Premium is $20, Team $30 per user, Business $100 per user, all metered by token consumption.
It produces the cleanest React and Next.js interface code of anything in the category, deploys to Vercel in a click, and generates standard code you can export to GitHub or run locally with no lock-in whatsoever. The trade-off is scope. The 2026 version added a full-stack sandbox, a git panel and database integrations, but full-stack work is still its weaker half. Hosting is also a separate Vercel bill, which people forget when comparing sticker prices.
Mobile is a different market
Almost everything above builds for the browser. Rork is the exception, generating React Native and Expo apps from $25 a month, with Rork Max producing native Swift apps for iPhone, iPad, Watch, TV and Vision Pro. Note the gap in that pricing table: native iOS is not on the $25 plan, it is on the $200 one. Emergent covers web and mobile from a single codebase through Capacitor. FlutterFlow remains the choice if you want exportable Flutter code and can absorb the learning curve.
Google’s answer sits at two levels. AI Studio now handles prototyping with one-click export into Antigravity for agentic multi-file development, while Opal covers no-code AI mini-apps in more than 160 countries. Opal is a demo and prototyping tool, though. No code export, no database, no service level agreement.
The part that does not show up in the demo
Escape scanned more than 5,600 publicly reachable vibe-coded applications and found over 2,000 vulnerabilities, more than 400 exposed secrets, and 175 instances of personally identifiable information including medical records, IBANs, phone numbers and email addresses. A RedAccess audit in May 2026 counted roughly 380,000 publicly exposed applications built on Lovable, Replit, Base44 and Netlify, with about 5,000 of them carrying sensitive corporate data. Base44 itself had an authentication flaw that let anyone register a verified account on a private application using only the publicly visible app ID; Wiz reported it, and it was patched within 24 hours with no evidence of exploitation.
The platforms respond that a public URL reflects a setting the user chose. That is technically true and practically useless, because the people building on these tools are precisely the people least likely to know what the setting does. A platform that defaults to private and makes you opt into exposure is safer than one that defaults open and hopes you opt into protection.
How to choose
| Platform | Strongest at | Cost signal |
|---|---|---|
| Lovable | Interface polish, non-technical founders | Pro $25, credits plus metered hosting |
| Base44 | Fastest prompt to live app | $16 to $160, dual credits, no rollover |
| Bolt.new | Speed, portability, bolt.diy escape hatch | Pro $25 for 10M tokens, one-month rollover |
| Replit | Shell, git, deployment control, mobile | Core $25 plus effort-based burn |
| v0 | Front-end quality, zero lock-in | Premium $20 plus a separate Vercel bill |
| Rork | Native iOS and Android | $25 to start, Swift output at $200 |
If you are validating an idea that may never see a second user, the choice barely matters. Pick whichever free tier you can stomach and move.
If the thing might survive, narrow it to two questions. Can you export the code and run it somewhere else? Does the platform default to private? Bolt and v0 pass both cleanly. Base44 fails the first outright. Replit passes both and still hands you enough rope to delete your own database, which one founder already did.