Disclosure: this article contains affiliate links. If you subscribe through one, we may earn a commission at no extra cost to you. It does not influence our recommendations — how this works.
Five tools now claim the same thing: describe an application, get a working application. They are not interchangeable. The differences that matter are not the ones in the marketing — they are portability, how you get billed, and how much of the decision-making you hand over.
Here is what each one actually is, and how to pick.
The short version
| Tool | Best for | Code export | Billing |
|---|---|---|---|
| Lovable | Beginners who want to keep their code | Full, via GitHub | Credits |
| Bolt | Control, terminal access, mobile apps | Full, via GitHub | Tokens |
| v0 | Teams already on Next.js and Vercel | Full, via GitHub | Credits from tokens |
| Base44 | Fastest path to a managed backend | Frontend only | Message + runtime credits |
| Replit Agent | Long autonomous builds | Platform-bound | Effort-based |
Lovable
Lovable was founded in Stockholm in December 2023 by Anton Osika and Fabian Hedin, growing out of the open-source gpt-engineer project. It generates React, TypeScript, and Tailwind, integrates natively with Supabase for database and auth, and deploys in one click.
Its defining choice is opinionation. You get one stack, well-trodden, and the platform does not ask you to make infrastructure decisions. Unlike the other managed option in this list, though, the code is genuinely yours: full export to GitHub, and Supabase is a service you own independently of Lovable.
Billing is in credits, metered per message in Plan Mode and by task complexity in Default Mode. Lovable renders its prices client-side and has changed packaging more than once, so check the pricing page rather than any figure quoted secondhand.
Choose it if you want a working app without touching a terminal, and you want to be able to leave.
Bolt
Bolt is StackBlitz's builder, launched in October 2024. Its technical bet is unusual: the entire development environment runs client-side, in a WebContainer inside your browser tab. The filesystem, the Node server, the package manager, the terminal, the console — all of it, local to you.
That has two consequences. The feedback loop is very fast, because nothing round-trips to a remote container. And you have a real terminal, which means Bolt is the one tool here that does not feel like it is hiding the computer from you.
Bolt is also the only one that produces genuine mobile apps: through its Expo integration it generates real React Native source, publishable to the App Store and Google Play. Publishing is not fully in-browser — EAS needs a local Node install and platform developer accounts — but the output is a real app, not a wrapped website.
Billing is by token: 1M per month free with a 300K daily cap, $25/month for Pro with 10M tokens and rollover.
Choose it if you want control, or you want a mobile app.
v0
v0 started as Vercel's UI component generator. It is now an agentic app builder, a shift Vercel marked by renaming it from v0.dev to v0.app in August 2025. The February 2026 release added a sandbox runtime, Git branches and pull requests, deploy-on-merge, and enterprise database integrations.
The agent plans its work, decomposes it into tasks, searches the web, inspects the running site, and fixes its own errors. Output is Next.js, React, Tailwind, and shadcn/ui.
Everything good about v0 follows from its coupling to the Vercel ecosystem, and so does everything constraining. If your team already ships Next.js to Vercel, nothing else here fits so cleanly. If you do not, you are adopting a stack along with a tool. Note also that authentication is not built in — you scaffold it with Auth.js or Clerk.
There is a Platform API, in public beta since July 2025, which exposes the whole generation pipeline over REST. Nothing else in this list offers that.
Choose it if you are on Next.js, or you want to build on top of a generation API.
Base44
Base44 is the most interesting story and the most consequential trade-off. Maor Shlomo built it alone, bootstrapped on his own money, starting as a side project in late 2024. Wix acquired it in June 2025 for approximately $80 million in cash plus earn-outs through 2029, and committed to running it as a distinct product — which, a year on, it still is.
The product bundles everything: a MongoDB-compatible database, authentication with social sign-in and custom SSO, Deno serverless functions, realtime subscriptions, row- and field-level authorization. You never configure infrastructure because there is no infrastructure to configure.
The catch is portability. GitHub export covers the frontend. The backend — queries, auth, business logic, endpoints — stays bound to the Base44 SDK. Leaving means rewriting it. This is reported consistently by independent reviewers rather than documented by Base44, and it is the thing to weigh before you build something you care about.
Watch the billing model too. Base44 charges message credits while you build and integration credits at runtime, every time your live app sends an email, generates an image, or calls an LLM. A successful app keeps costing you. Neither credit type rolls over.
Choose it if speed to a working backend matters more than the ability to migrate away from it.
Replit Agent
Replit Agent is the most autonomous option, and it should not be confused with Replit AI, which is the older autocomplete-and-chat assistant. Agent builds; the assistant helps you build.
Agent 3, announced September 2025, runs unattended for up to 200 minutes. It writes code, provisions a PostgreSQL database and auth, opens the running app in a browser, exercises it, and repairs what it finds broken — repeatedly, without being asked. It can also build other agents: Slack bots, Telegram bots, scheduled automations.
Pricing is effort-based: a checkpoint costs in proportion to the time and compute the request consumed. Trivial requests cost under $0.25; complex ones cost more. You find out afterwards. Turn on usage alerts and a hard budget cap before your first long run.
One episode deserves mention, because it is the clearest illustration of what autonomy costs. In July 2025, during a multi-day experiment by SaaStr founder Jason Lemkin, the agent deleted a live production database despite an explicit code freeze, destroying records for roughly 1,200 executives and 1,190 companies. Replit's CEO called it "unacceptable and should never be possible," and the company added automatic development-production database separation, better rollback, and a planning-only mode. The incident predates Agent 3 and the December 2025 database changes. It remains a useful reminder that an agent with 200 minutes of autonomy and database credentials is a category of tool that deserves a budget cap and a backup.
Choose it if you want the machine to do the whole loop, and you are prepared to supervise it anyway.
How to actually choose
Start with portability, because it is the only decision you cannot reverse cheaply. If you might ever want to leave, that rules out Base44's backend and, in practice, Replit's platform. Lovable, Bolt, and v0 all hand you the code.
Then look at the billing model against your usage shape. Base44's runtime integration credits punish success. Replit's effort-based pricing punishes complexity you did not anticipate. Bolt's token billing punishes long chat histories, because context is re-sent. Nobody here charges a predictable flat rate, and anyone who tells you they can forecast their monthly bill has not built anything complicated yet.
Then match the tool to the stack you already have. On Next.js and Vercel, v0 is the obvious answer. Building a mobile app, Bolt is the only real option. Wanting no infrastructure decisions at all and willing to accept the lock-in, Base44.
Otherwise, Lovable is the reasonable default for most people building a web app: opinionated enough to be fast, open enough to leave.
The cheapest way to decide is to build the same small application in two of them. All five have free plans. An afternoon spent doing that will tell you more than any comparison table, including this one.
Related reading
- Lovable vs Bolt — the terminal, mobile apps, and token billing
- Lovable vs v0 — the Vercel gravity well and the authentication gap
- Lovable — full profile, pricing, and limitations
- Bolt — WebContainers, Expo, and token billing
- v0 — the Vercel stack and the Platform API
- Base44 — the managed backend and its price
- Replit Agent — autonomy, effort pricing, and the database incident
- Cursor — if you would rather write the code yourself