Lovable for the fastest polished first version with a backend included. Bolt if you iterate in long sessions and want a large monthly token allowance. v0 for React and Next.js interfaces that deploy straight to Vercel. Replit Agent for full-stack apps with a database and the most built-in infrastructure. Claude Code or Cursor once the project is a real codebase that needs tests, reviews and a team, with Claude Code the stronger fit for agent-led work and Cursor for people who want to stay in an editor.
I run Space & Story, a Toronto studio that builds web and AI products for clients, and I won the Grand Prize in Lovable's Shipped season 1, so read the Lovable section knowing I have a good history with it. This site was built with Claude Code. For the other four, the comparison is from their published specs and documentation, not a head-to-head test. Prices and plan features below come from each vendor's pricing page or plan documentation, opened on October 31, 2026, and they move often, so each row links to its source.
The short version
| Tool | Best for | Weakest at | Code ownership |
|---|---|---|---|
| Lovable | A polished full-stack first version from a description | Long-tail logic once the app grows | Git sync on all plans |
| Bolt | Long iterative sessions in the browser | Larger apps and team workflows | Not stated on the pricing page |
| v0 | React and Next.js interfaces on Vercel | Back-end-heavy work | GitHub sync on all plans |
| Replit Agent | Apps that need a database and hosting in one place | Fine control over the stack | Code lives in your Replit workspace |
| Claude Code | Real repositories, multi-file changes, tests, CI | A first visual demo | Your repo, your machine or cloud |
| Cursor | Developers who want an AI-first editor | People who don't want to read code | Your repo |
Pricing, checked October 31, 2026
| Tool | Free | Main paid plan | Higher tiers | Source |
|---|---|---|---|---|
| Lovable | 5 daily build credits (max 30 a month); no custom domain or code editing | Pro $25/month for 100 credits ($21/month billed annually) | Business $50/month ($42 annual) with SSO; Enterprise by contract | Lovable docs: plans |
| Bolt | 1M tokens a month, 300K daily limit, hosting, Bolt branding | Pro $25/month, 10M tokens, rollover, custom domains | Teams $30 per member/month; Enterprise custom | bolt.new/pricing |
| v0 | 7 messages a day; deploy to Vercel, GitHub sync | Plus: the page shows both $30 and $90 per user/month; $30 of monthly credits plus $2 daily | Business $100/user/month; Enterprise custom | v0.app/pricing |
| Replit | Not shown in our fetch | Core $20/month ($18 annual), $20 toward premium models | Pro $100/month ($90 annual), up to 15 collaborators; Enterprise custom | replit.com/pricing |
| Claude Code | Not on the Free plan | Included with Claude Pro, $20/month ($17 annual) | Max $100, Ultra $200 per month; Team Premium $125/seat monthly | claude.com/pricing |
| Cursor | Limited agent and chat requests | Premium $30/month, cloud agents, AI code review, third-party models at API list price | Super $100, Ultra $200; Teams from $30/user/month | cursor.com/pricing |
USD list prices before tax. Lovable charges per credit rather than per prompt; its own examples range from 0.5 credits for "make the button gray" to 1.7 for a landing page with generated images. Cursor's plans are now bundled with xAI's Grok products: its $8 Plus plan includes Grok models only, which is why we list Premium as the main plan for coding.
The headline prices are close; what differs is the unit. Lovable and v0 meter credits, Bolt meters tokens, and Replit, Claude Code and Cursor sell a usage allowance against model costs. A heavy week of iteration can cost several times a light one on any of them, so the useful comparison is your own first project, not the list price.
Lovable
Lovable turns a description into a running web app with a backend, auth and hosting, and it's the best of the group at making the first version look finished. Every plan syncs to GitHub, GitLab or Bitbucket, so you're never locked in, and Pro unlocks custom domains and direct code editing. Where it struggles is the same place every prompt-to-app tool does: as an app grows, each fix is a new prompt into a codebase you didn't write, and credits go faster on the tenth fix to the same screen than on the first build. Pick it for MVPs, internal tools and anything you need to put in front of users this week.
Bolt
Bolt runs a full development environment in the browser and meters work in tokens, with 10M a month on Pro and unused tokens rolling over. That suits people who work in long sessions and want to see and tweak the code as they go. Hosting is included on every plan and custom domains start at Pro. Its pricing page says little about exporting code or how databases are provisioned beyond "unlimited databases", so test that path early if you plan to move the project elsewhere.
v0
v0 is Vercel's builder and it's at its best on React and Next.js interfaces: components, dashboards, marketing pages that deploy to Vercel in one step, with GitHub sync even on the free plan. The free tier is tight at seven messages a day. Its paid pricing was ambiguous when we checked, with both $30 and $90 shown for Plus, so confirm on the live page. Choose it if your stack is already Vercel and you mostly need front-end work done well.
Replit Agent
Replit packages the agent with the workspace, database and hosting, so a full-stack app with persistent data needs no other accounts. Core is $20 a month and Pro $100 with up to 15 collaborators and 28 days of database rollback. The trade-off is that you work inside Replit's environment and conventions; if you want a specific framework or host, you'll spend effort moving off it later.
Claude Code
Claude Code is a coding agent that works in a real repository from your terminal, IDE, desktop app or the cloud. It reads the codebase, makes multi-file changes, runs your tests and commits, and it can be extended with skills and MCP servers like the rest of Claude. It has no hosting or visual editor, so the first demo is slower than Lovable's. It earns its place once there's real complexity: several services, a test suite, CI, a team reviewing pull requests. It comes with every paid Claude plan from Pro at $20 a month, with Max and Ultra for heavier use.
Cursor
Cursor is an AI-first code editor with an agent, cloud agents that work in the background, and AI code review on Premium and above. It's the natural pick for developers who want AI inside the editor they already think in, and less so for anyone who doesn't want to read code. Its plans have changed shape: the cheapest paid tier is now a Grok bundle, and for coding with a choice of models the relevant plan is Premium at $30 a month, with third-party models billed at API list price.
Who should pick which
| You are | Start with | Then |
|---|---|---|
| A founder testing an idea with users this month | Lovable or Bolt | Export to GitHub once the idea holds |
| A designer or front-end team on Vercel | v0 | Hand the repo to a coding agent for back-end work |
| A solo builder who wants database and hosting in one place | Replit Agent | Stay while it fits; plan a move if you need a specific stack |
| A developer or team with an existing codebase | Claude Code or Cursor | Add skills for your conventions and CI review |
| An operator automating internal work | Probably none of these | Run the task through the decision tool; most internal jobs are a skill or a connector, not an app |
When you've outgrown them
The builders are excellent at the part of software that's visible. They get harder to steer when the work becomes invisible: data models with history, roles and permissions, payments, integrations that must not fail quietly, and the slow accumulation of edge cases where a fix in one place breaks two others. The usual sign is a credit bill that keeps growing while the app stops improving.
At that point you have two good options. Move the code into a real repository and keep going with a coding agent, adding tests before you add features. Or, if the product matters and nobody on the team builds software for a living, bring in people who do. That's the work my studio, Space & Story, does, so weigh that suggestion knowing I have an interest in it; the first option is often enough.
And before any of this, check that it should be an app at all. Plenty of things people build in a weekend on these tools would have been a skill and a connector, which is the subject of the explainer.
Questions people ask
What is the best AI app builder for non-developers?
Lovable or Bolt. Both turn a description into a running web app with hosting included, and neither expects you to open a code editor. Lovable is stronger on a polished first version and its built-in backend; Bolt's token-based plans suit people who iterate in long sessions.
Is Claude Code an app builder?
It's a coding agent rather than a builder. It works in a real repository on your machine or in the cloud, runs commands and tests, and has no hosting or visual editor of its own. That makes it slower to a first demo than Lovable or Bolt and much better once the project has real complexity.
Can I move a Lovable or Bolt project to my own code later?
Lovable syncs to GitHub, GitLab or Bitbucket on every plan, including Free, so the code is yours to take. Bolt's pricing page doesn't say how export works, so check before you commit a large project.
When should I stop using an AI app builder?
When the work stops being about screens and starts being about data, permissions and edge cases. Signs include several user roles, payments, integrations that must not fail, and a growing list of fixes that break other fixes. That's the point to move into a real repository with a coding agent, or to bring in people who build software for a living.