Vybes is now valued at $10M.Read more

AI & strategy

Replit Alternatives for an App With Real Users

The right Replit alternative depends on why you are leaving. Agent costs, hosting limits, data ownership, code quality, mobile and security each point to a different move, and for some apps the right move is to stay.

Nate Laquis13 min read

The Short Answer on Replit Alternatives

For an app with real users, the best Replit alternatives are a GitHub repository and hosting that you own, with a developer working in the code using Cursor, Claude Code or Codex. That means Vercel, Railway, Render or Fly for the app and Supabase or Neon for Postgres. Lovable, Bolt, v0 and Base44 are the right swap only when the app is still a prototype and your one complaint is the Agent bill.

There are three kinds of move. You can switch to another builder. You can keep the code and put a developer on it with an AI coding tool. Or you can take the code and the data off Replit entirely. The second and third usually happen together.

I checked every vendor's pricing page and docs on November 18, 2026. These numbers change every few months, so treat them as a snapshot and look again before you pay.

Six Reasons Founders Leave Replit

Name your reason before you compare tools. Each one sends you somewhere different, and most of them are not fixed by another builder.

  • The Agent bill. Replit prices Agent by effort, so you cannot know what a request costs until it is done. A cheaper builder can help here. So can staying and changing how you work, which is covered near the end.
  • Hosting limits. Replit's docs say all published apps are hosted in the United States, and EU hosting is something Enterprise customers ask sales for. If a customer needs their data in Europe, or you want more control over how the app runs, you want your own hosting.
  • The database. You want backups you control, a connection your own tools can use, and a clear answer to who holds the data. Your own Postgres account answers all of it.
  • Code quality. Each fix breaks something older, and Agent spends credits circling the same bug. A fresh builder gives you a clean start and the same problem a few months on. This one needs a person who reads the code.
  • Native mobile. Replit builds Expo apps and can get one to the App Store. Its docs say Google Play publishing is not supported yet.
  • Security and due diligence. A customer sends a security questionnaire, or an investor asks who maintains this. Scanners help with the first. The second needs a repository in the company's name and a named person responsible for it.

Only the first reason is solved by switching builders. Hosting and data point to your own accounts. The last three point to a developer.

What Replit Charges Once People Use the App

Start with what you pay now, because a lot of founders have never added it up.

Core is $20 a month, or $18 on annual billing, and includes $20 of credits. Pro is $100 a month, or $90 on annual billing, with $100 of credits, up to 15 collaborators and database rollback up to 28 days. Pro also sells larger credit tiers that run up to $2,000 a month.

Agent draws from those credits. Replit calls it effort-based pricing: a small bug fix costs less than a full feature, and each request is billed as one checkpoint when the work is finished. Planning is billed too in the Power and Max modes, even when no code changes. You see the cost of a checkpoint after the fact by hovering over the usage icon. That is the unpredictable part. A request that looks small can turn into a long run, and you learn the price afterwards.

Hosting comes out of the same credits. Replit's docs list these prices:

  • Autoscale: $2 a month base, plus $0.60 per million compute units and $0.40 per million requests.
  • Reserved VM: $15 a month for a shared half vCPU with 2 GB of memory. Dedicated machines are $35, $50 and $130 a month, the largest with 4 vCPU and 16 GB.
  • Static: free to host, with data transfer at $0.05 per GB.
  • Scheduled jobs: $2 a month base plus compute.

The production database is billed on compute and storage. It goes idle after five minutes without activity, and storage keeps billing while it sleeps.

So one pool of credits pays for building, hosting and the database. A heavy week of Agent work eats the same balance that keeps the app online, and after the credits run out you pay by usage. Set a budget and an alert under Settings, then Billing, today. It takes two minutes and it is the cheapest fix in this article.

Two developers working at laptops side by side in a bright office

Lovable, Bolt, v0 and Base44

These four do the job Replit Agent does. You describe the app and the tool writes it, runs it and gives you a link. Here is what each pricing page listed.

Lovable

Free gives 5 build credits a day, up to 30 a month. Pro is $25 a month with 100 credits and Business is $50. A plan buys a fixed number of credits, which is easier to budget than effort-based checkpoints. Pick it if you want a web app with a tidy front end and you never needed Replit's scheduled jobs or VMs.

Bolt

Free with 1 million tokens a month and a 300,000 daily limit. Pro is $25 a month, starts at 10 million tokens and has no daily limit. Teams is $30 per member. Tokens track how much code the model reads and writes, so a big project costs more per prompt than a small one.

v0

Vercel's builder. The free plan has a limit of 7 messages a day and includes GitHub sync and deploys to Vercel. Plus is listed at $30 per user a month with $30 of credits, and Business at $100 per user. Pick v0 when the screens are the hard part.

Base44

Free with 25 message credits a month. Paid plans are $16, $40, $80 and $160 a month on annual billing, for 100, 250, 500 and 1,200 message credits. I would look at it for an internal tool.

The catch with all four

Whatever a builder can import, it will not bring your Replit database, your Replit Auth users or your deployment settings. In practice you rebuild from prompts with the old app as the reference. A prototype with a few screens takes an afternoon. An app with users means a full rebuild and a data migration, and at the end nobody has read the new code either.

Switch builders for a bill you can predict. For any of the other five reasons, skip this category. Our Lovable alternatives post covers the same question from the other direction, and the conclusion matches.

Cursor, Claude Code or Codex in the Hands of a Developer

This is the move I would look at first if people already use your product. Cursor, Claude Code and OpenAI Codex work inside an ordinary code project. They read the whole repository, change many files in one pass, run the tests and show what they changed.

They also work on the code you already have. Push the Replit project to GitHub, a developer clones it, and work continues from where Agent stopped. You throw nothing away on the first day.

What each vendor's page listed when I checked:

  • Claude Code: included in Claude Pro at $20 a month, or $17 a month on the annual plan. Max is $100 and Ultra is $200.
  • Codex: included with ChatGPT plans, with limited access on Free and on Go at $8. Plus is $20 a month and Pro starts at $100.
  • Cursor: Plus is $8 a month and runs Grok models only. Premium is $30 and opens up other companies' models, billed separately at API list price. Above that are tiers at $100 and $200.

The subscription is a rounding error next to the developer's time, and the developer's judgment is what you are paying for. An agent makes dozens of decisions per task: where the permission check goes, whether a payment is confirmed on the server, which columns get an index. On Replit you were the only reviewer of those decisions, and you were not reading the code. Here changes arrive as pull requests that someone reads before they go live.

A founder who does not code can open Claude Code or Codex and get further than you would expect. Do that to learn, and for small fixes. I would not do it alone on an app that holds customer data. Our Replit vs Cursor comparison goes deeper on how the two ways of working differ.

Getting Your Code and Data Out of Replit

Replit has a Git pane and a Shell, so the code leaves easily. The data and the sign-in take the real work. Here is the order I would do it in.

  1. Push the code to GitHub. Open Tools, then Git, and connect a repository under the company's GitHub account. From the Shell it is a plain git remote and a personal access token. There is also a Download as zip option in the file tree menu. Do this now even if you stay. It is your backup.
  2. Write down what the zip leaves out. Replit's docs say secrets, the database and deployment configuration are not included when a project is copied. Copy every value out of Secrets by hand, and rotate any key that was ever pasted into a chat.
  3. Dump the production database. It is PostgreSQL, and the docs say you can connect to it from any PostgreSQL-compatible client with the connection string in the production database's Settings tab. A developer runs a standard dump and restores it into Supabase or Neon. The development database is different: the docs say it cannot be reached by external database viewers, so do not plan around it.
  4. Download uploaded files. App Storage runs on Google Cloud Storage. Files can be downloaded one at a time from the Objects view, and a short script handles a large bucket.
  5. Replace Replit Auth. Its login page is Replit-branded and its accounts are tied to Replit. Replit's docs point to Clerk for accounts that are not, and they include a guide for having Agent migrate an eligible app from Replit Auth to Clerk. Do that migration while you are still on Replit, test it with a new account, then move.
  6. Stand up the new host. Deploy to a test domain, run both copies side by side with real accounts, then switch the domain.

On the hosting side, Vercel's Pro plan is $20 a month and suits a Next.js or React front end. Railway is $5 a month on Hobby with $5 of usage included, and $20 on Pro. Render's smallest paid web service is $7 a month. Fly bills machines by the second, with the smallest listed at $2.19 a month, and has no permanent free tier. For Postgres, Supabase Pro starts at $25 a month, and its free plan pauses a project after a week of inactivity, so do not run production on it. Neon has a free plan and paid plans billed purely on usage.

My estimate for a young product is $25 to $75 a month all in, and the number barely moves from month to month.

For a small app with a few hundred users, my estimate is one to three weeks of a developer's time. Sign-in takes the longest, and every outside service you connected adds a day of keys and callback URLs.

Close-up of Python code with line numbers on a dark laptop screen

Native Mobile, Security and Due Diligence

These three reasons all end with a person doing the work, whichever tool you pick.

Native mobile

Replit's Agent builds mobile apps with Expo and React Native. For iPhone you publish from the editor to TestFlight and promote the build in App Store Connect, with your own Apple Developer Program membership. The docs are plain about the rest: publishing to Google Play is not yet supported, and neither are EAS CLI and EAS Build, the standard Expo build tools. That is fine for an iPhone pilot. It is thin for a consumer app that needs both stores, video, payments or identity checks.

The output is a standard Expo project, so it moves to your own repo like any other code. For a sense of what a full build looks like, see Vybes. We built it as a React Native app for iOS and Android on a Node.js backend, with real-time WebRTC video calls and document verification. It went from concept to the App Store in eight weeks and holds a 4.8 rating. A product like that needs engineers on the native layer from the first week.

Security

Replit has more security tooling than founders tend to know about. Changes get a lightweight check as Agent builds, a scan runs before you publish, dependency scans are free, and paid plans add Security Agent to audit the codebase. Turn all of it on. Then read Replit's own wording: the build-time check is not a full codebase review, and you are told to review each finding yourself.

A scanner finds known patterns. It does not ask whether user A should be able to load user B's invoice by changing a number in the URL. Someone has to read the endpoints and ask that question about each one.

Due diligence

An investor or an acquirer wants four things: a repository the company owns, hosting and database accounts in the company's name, a stack an engineer can be hired onto, and a named person or team who maintains it. Replit can give you the first if you connect GitHub. The others come from the move described above and a signed contract with whoever takes the code on.

Hand holding a phone that shows a mobile app dashboard, with a laptop and monitor behind it

Reasons to Stay on Replit

Some of you should not move at all.

Stay if you do not yet know whether anyone wants the product. Replit is still the fastest way I know to put a working backend in front of a buyer, and speed is all that matters at that stage.

Stay if the app holds nothing sensitive and takes no payments. A team dashboard or a waitlist can be down for a morning and nobody is hurt.

Stay if the app works and the only complaint is the bill. Four habits bring it down:

  • Use Free Mode for everyday questions. Replit asks before any paid action starts.
  • Send one bug or one screen per request. Effort-based pricing punishes vague, sprawling prompts.
  • Roll back to a checkpoint when a change made things worse. Prompting forward through the damage costs more.
  • Check your deployment type. An app with a handful of users on a dedicated Reserved VM is paying for a machine it does not need.

There is a middle route too. With the repo on GitHub, a developer can fix the one stubborn problem in a pull request while you keep building in Replit. A few hours a month of that often costs less than the credits the loop was eating.

If the app is misbehaving and you cannot tell which side of the line you are on, our guide on when to hire a developer to fix a Replit app lists the specific signs.

Cases Where You Should Not Hire an Agency

We sell this work, so read the next four paragraphs knowing that. In each of these cases I would tell you to keep your money.

Nobody has used the app yet. Everything that makes Replit risky comes from real people depending on the app. Until they do, the fastest tool is the right tool.

The budget is a few thousand dollars. A takeover by a team costs more than that, and an agency that says otherwise will cut something you will not notice until later. Spend the money on a written code audit, or on a good freelancer for a week of targeted fixes. Either one leaves you with something you can use.

The only problem is the Agent bill. That is a prompting and plan problem. Fix it with the habits above, or try a builder with flatter pricing.

The code just looks ugly. Agent writes long files and repeats itself. If the app is safe and changes still go through without breaking older features, leave it alone.

Hire a team once customers pay you, or the app holds data people would be upset to see leaked, or a funding round with technical diligence is on the calendar. From then on a missed permission rule or a failed migration costs more than the work to prevent it. For market ranges on that work, see what it costs to rescue a vibe-coded MVP.

Which Move Fits Your App

  • A prototype with no users and an Agent bill that annoys you: stay and change how you prompt, or try Lovable or Bolt at $25 a month.
  • An internal tool for your own team: stay on Replit.
  • A front end that has to look sharp: v0.
  • A working web product where each change breaks something: push the code to GitHub and bring in a developer with Claude Code, Codex or Cursor. Hosting can stay on Replit for now.
  • Hosting costs you cannot predict, or a customer who needs EU data residency: your own hosting and your own Postgres.
  • Paying users and private data: an audit first, then the move to accounts the company owns.
  • An iPhone pilot: Replit's Expo path is enough.
  • A consumer app on both stores: take the Expo project to your own repo and put a React Native team on it.
  • An investor asking who maintains this: company-owned repo, company-owned hosting, and a named team on a contract.

Taking over Replit-built apps is what we do. We read the repository, tell you what to keep and what to rebuild, move the code and the data to accounts you own, and stay on as the team that maintains it. If your app has users and you have hit one of the six reasons above, tell us which one. Book a free strategy call

replit alternativesReplit Agent pricingexport Replit app to GitHubAI app buildervibe coded app rescue

Your Replit app, ready for real users.

Send the link. You get a free teardown of what breaks first and what it takes to fix it.

Get a free teardown