Lovable Pricing in 2026: What the Credit Model Hides About Real Project Cost

fuse-smo-martin-janecekWritten by Martin J.
Back to blog
A glowing orange line climbing steeply across a frosted glass dashboard panel, above a flat blue baseline that stays level and never follows it, on deep navy

Twenty-five dollars buys a month of Lovable, and a month is not what a project needs. The first build feels like a rounding error. The second arrives with a different shape, and nothing on the pricing page predicts the gap. So what are you buying?

The first app you build with an AI tool feels like a rounding error. Twenty-five dollars goes in, a working thing comes out, and you start telling people about it over lunch. Then the second one arrives with a different shape: more screens, a login, a database, and a client who changes their mind on Thursday. The bill at the end of that month is not twenty-five dollars. Nothing on the pricing page was wrong, and nothing on it predicted the difference. So what were you actually buying the first time, and why does the same $25 feel like a gift in March and a warning in November?

There is a number on Lovable's pricing page that most buyers read as the price of the product, and it is not the price of anything. It is the first rung of a selector. Behind it sit eleven tiers on Pro and eleven more on Business, and the unit those ladders count carries two different prices on the vendor's own website and documentation, both read on 18 November 2026. Your project does not care about either of them. It cares how many messages it takes to finish, and that number is set by a phase of the work that appears in no budget you have ever seen. A two-week internal tool can land inside one $25 month. A slightly larger one can cost more than the laptop it runs on. The plan was not the variable in either case.

The Plans

Lovable sells four tiers: Free, Pro, Business and Enterprise. The plan cards on lovable.dev/pricing did not render in any automated read on 18 November 2026, so the grid below comes from the vendor's own documentation at docs.lovable.dev, section "Subscription tiers and billing cycles", read the same day.

Plan

Monthly

Annual (as published)

Credits included

Free

$0

$0

5 build credits per day, capped at 30 per month

Pro

$25

$250 per year, shown as $21 per month

100 per month

Business

$50

$500 per year, shown as $42 per month

100 per month

Enterprise

custom

custom

volume based

Two things in that table matter more than the prices. Annual billing is two months free, which is what $250 against $25 a month works out to. There is also no trial on a paid tier, so you cannot test Pro for a week before the charge lands.

The $25 figure is the first rung of a selector rather than the price of your plan. Pro runs from 100 credits for $25 to 10,000 credits for $2,250 across eleven tiers, and Business runs the same eleven at exactly double up to 5,000 credits. Nothing is published above 10,000. Enterprise begins with a conversation.

If a third-party guide told you Pro is $25 for 100 credits, it was right. Those figures are not in dispute. The argument is only about what sits behind them.

Free and the paid tiers are not the same product with a bigger allowance. Free gets 5 build credits a day capped at 30 a month, with no top-ups, no custom domain, no code download and no rollover, which rules out anything you plan to keep.

A Credit Is Not a Request

Here is the vendor's definition, and it is worth reading twice:

A credit is the unit Lovable uses to measure and pay for usage across your workspace. Credits let you build apps, chat with Lovable, run deployed apps, and power AI features from one balance.

That one balance pays for five different things: building, chatting (including Slack and Telegram and each automation run), Cloud hosting with the built-in database and compute, model calls your deployed app makes, and managed connector requests such as Perplexity or Firecrawl. So your building and your running draw on the same number.

Two modes price differently. Plan mode charges one credit per message plus any subagent research, and it never modifies your code. Build mode charges by the complexity of the request and the work completed.

That pooling is recent, and the vendor dates it: credit history starts 8 March 2026, and run-credit history starts whenever a workspace moved over, 1 June 2026 at the earliest. Treat the merge as a project risk. A heavy week of iteration can quietly spend the credits that were keeping a live app's database up.

Now the part that should bother anyone building a budget. The same prompt carries two prices on two first-party pages, both read on 18 November 2026. "Build me a landing page, use images" costs 1.70 credits in the pricing page FAQ and 2.00 credits in the documentation. The three smaller published examples match to the decimal. The largest one does not.

That is 7.5 cents at Pro rates, so the money is not the issue. The issue is that the vendor's own site and docs disagree on the unit price of the unit your whole budget is counted in. Both are labeled illustrative, which is fair. It also means you cannot budget from the examples, and you need a model instead.

Three habits make the number predictable. Send one change per message, because Build mode prices the complexity of the request. Plan before you build, since Plan mode never modifies your code. And watch the pause: a single message can run up to 10 hours, and Lovable stops when it crosses your credit check-in level, 20 credits by default, which is $5 at Pro's rate.

A month is a calendar unit, and work does not arrive in calendar units. Forty small changes and three days on one broken login flow land on the same invoice. Budget the project instead, divide by the months you honestly expect it to take, and treat the result as a floor.

Costing a Project, Not a Month

A build does not move at one average rate. It passes through three phases with three different cost curves, and the third is where your project goes wrong.

Three groups of glowing glass columns inside one dark panel: a few tall blue bars, then a dense run of short ones, then the most and tallest of all in orange

Scaffolding is the first session: layout, pages, auth, data model, theme. Generation writes a lot at once, so you get your highest cost per message and your fewest messages in the same afternoon. The vendor's landing page example is the anchor here, at 1.70 to 2.00 credits for three generated images, a theme and five sections.

Iteration is everything after: copy, spacing, column order, an empty state, a field that should have been a dropdown. "Make the button gray" is 0.50. You pay for volume here, not for unit price, which is why iteration feels cheap message by message and expensive by the end of the month.

Debugging is the phase with no line in your budget. Something breaks, you ask for a fix, the fix breaks something else. The vendor charges for work completed, so a failed attempt and its repair are two billable messages, and their caveat adds that cost rises with codebase exploration, verification, browser checks and web search. Highest unit price and the most messages at once, which is the wrong way round for anyone hoping the hard part comes first.

What follows is a model I derived from the published per-message examples. Nobody measured it, so treat the shape as more reliable than the decimals. At least one calculator on the web already turns credits into project ranges, and its numbers are not wrong either. What none of them do is split the bill by phase, which is the part that tells you whether you should have moved up a tier two months ago.

The build is an internal dashboard for a five-person team. Login and roles, one database table of records, a list view with filters, a detail view, a form to add and edit, and a CSV export. No payments, no customer facing traffic, no AI features inside the app. totalum.app publishes an estimate that brackets the same shape at 80 to 150 build credits, and it flags that number as its own estimate, not a measurement.

  • Scaffolding, five messages anchored to the vendor's examples: about 8 credits.
  • Iteration, 50 small changes averaging 0.70: about 35 credits.
  • Debugging, two genuinely stuck problems at five messages each, priced at the expensive end: about 16 credits.

Total, about 59 credits. At Pro's $0.25 that is roughly $15 of credits inside a $25 month that already includes 100, so you finish the tool with 41 credits to spare.

That is the honest answer, and it is less alarming than most writing on this subject: a small internal tool is cheap. The useful part is where the money sits. 51 of the 59 credits, 86 percent, are iteration and debugging. Most of the bill goes on changing a working thing, not on producing the first version of it.

And the debugging phase is not a thought experiment. zite.com, reading through Trustpilot, Reddit and G2, reports that unpredictable credit consumption is the most common complaint about Lovable. I have not read those reviews myself, so treat that as second hand. The mechanism is first party: a failed attempt and its fix each cost a full message.

Hosting is the one line that does not bite here. The 20-credit monthly Cloud grant covers it, and the docs say running an app costs nothing for most users with smaller or new apps. The hidden cost sits entirely in the debugging phase.

So why pay at all? Because the Free tier gives you 30 build credits a month, and 59 divided by 30 is two calendar months. Two months is a deadline, and a tight one. Free also cannot take a custom domain or let you download the code, which matters the moment the tool has a user who is not you.

The switch point is the arithmetic nobody publishes. Top-ups on Pro cost $0.30 a credit, so 84 of them is $25.20, which is more than the next tier up costs over the entry rung. Above 83 top-up credits in a month, upgrading is cheaper on the same money, and Business lands on the identical number. The switch you should have made is the month you bought your second pack.

Top-Ups and the Real Ceiling

Top-ups are the release valve, and they are priced like one. Pro pays $15 for 50 credits, Business $30 for 50, which is $0.30 and $0.60 a credit. Both sit exactly 20 percent above the plan rate, and the price does not change on annual billing, so a top-up costs you the same in month one and month twelve. Packs run from 50 to 10,000 credits. There are none on Free: the docs say to upgrade to a paid tier to unlock them.

A long ascending staircase of translucent glass rungs climbing out of blue light into orange, forking into two rungs at different heights near the top

The ladder itself barely discounts. Pro is $0.25 a credit from 100 up to 800, slips to $0.245 at 1,200, and reaches $0.225 at 5,000. Business does the same, from $0.50 to $0.43. A bigger tier below 1,200 credits buys a bigger bucket at the same rate, so it saves you a top-up and nothing else.

Auto top-up ships with three defaults worth knowing. 100 credits per refill, a trigger at 25 credits remaining, and a monthly spend limit of 400 credits. The vendor suggests raising the trigger to 50 or 100 for a busy app, which is a quiet admission that 25 is too close to empty.

The ceiling is public up to a point. 10,000 credits costs $2,250 on Pro and $4,300 on Business, which is the last number you can actually plan against. Above it nothing is published, and Enterprise is volume priced. Do not let anyone hand you a crossover figure: there is no public number to compute one from.

Annual billing buys you two months, and it buys nothing else. You pay ten months for twelve, the credits still arrive monthly, and they still expire. Monthly plan credits die two months after issue, annual plan credits one month after the annual period ends, top-ups twelve months after purchase. None of it is refundable, so a three-week build that finishes in week two leaves a use-it-or-lose-it balance.

What the Plan Price Does Not Cover

Start with what the paid entry tier buys that Free does not, because credits are the smaller half of it. Custom domains, removal of the Edit with Lovable badge, editing and downloading code, design systems, credit rollover, top-ups, user roles and per member credit limits, and email support. Several of those separate a demo from something you can put in front of a client.

Business doubles the credit price and adds no run capacity. The vendor says so outright: running your app and the AI features built into it cost the same on Business as on Pro. What the extra $25 per 100 credits buys is governance. SSO, role-based access, the security center, commit attribution, design templates, workspace publish, the API, and workspace data excluded from AI training by default. That is a procurement purchase.

Enterprise adds the compliance pieces your security review will ask about: SCIM, audit logs, organizations, data exports, scheduled security scans, EU inference, publishing and sharing controls, dedicated support and onboarding.

Seats never move the bill. Every tier allows unlimited members, so inviting your whole team costs nothing. Usage is the only lever.

What no plan covers is the part you are still doing. Nobody in this price list decides what the tool should do, writes the copy inside it, or gets anyone to visit. You still own the data model, the naming and every decision about what the screen should say.

An app that generates signups still needs a campaign behind it, and that is a separate purchase, which is why the AI campaign builder sits in a different budget line from the build.

Lovable Against the Rest of the Stack

Lovable is a build layer. It turns a description into a working web app and does that one job. The rest of a stack is assembled around it, and the shape of that assembly is what vibe marketing describes.

Two neighbors are worth naming, and neither is a substitute. n8n for marketing handles the plumbing between tools: triggers, schedules, handoffs. Claude Code for marketing works on a codebase you already own instead of generating a new one. Zapier pricing answers a similar question for a different product class, automation rather than generation.

The line that decides where each tool sits is the subject of what is agentic marketing: a system that decides what to do next versus a tool that waits for a prompt. Lovable is the second kind, and it is very good at the second kind. If you want the wider map, agentic marketing tools covers the tool against tool comparisons that this page deliberately leaves alone.

Who Should Pay for Which Plan

Solo builder, internal tools only: Pro at 100 credits for $25, billed monthly. You will finish a dashboard like the one above inside a single month, and you should hold off on annual billing until you have three months of your own numbers. Annual saves two months, not two months plus a cushion.

A team shipping client work: Pro at 400 or 800 credits, and pick the tier from what you actually used last month rather than what you hope to use. There is no volume discount below 1,200 credits. 400 costs $100 and 800 costs $200, the same $0.25 a credit. Above 83 top-up credits in a month, move up a tier instead.

An agency or a company with a security review: Business, bought for SSO, role-based access and audit. At the same credit count it costs exactly double, and it runs your app at the same price. If nobody in procurement has asked for single sign-on, you are paying for a feature you will not use.

Two edge cases. Students and teachers get 50 percent off Pro at 100 credits billed monthly, so $12.50 for up to 12 months, and verification is only available to workspaces on Free. And anyone whose project fits inside 60 credits a month should stay on the smallest paid tier, because 60 credits fits inside the 100 the $25 plan already includes and there is nothing to top up.

One profile should not pay at all yet. If you are testing whether an idea deserves to exist, stay on Free for two weeks, build the ugliest version that answers the question, and spend nothing. The moment the answer is yes, you are buying a custom domain and a code download, and that is the $25 talking rather than the credits.

And one profile should not be here at all. If what you need is a marketing site or a landing page, a tool that generates an application is the wrong purchase, even though it can technically produce a page.

Conclusion

The number on the pricing page is the smallest thing you will learn about Lovable's cost. Messages set the bill, each message is priced by the work it does, and that work clusters in the two phases nobody budgets for: the iteration that makes a tool usable and the debugging that makes it stop breaking.

So budget the project, not the month. Count the changes you expect, add room for two stuck problems, and check whether the total crosses 83 top-up credits before you buy a pack.

What I keep coming back to is that this is a cheap way to build something small and an expensive way to build something badly. Which of the two you get is decided before you pick a tier, in how honestly you scoped the thing and how quickly you notice that a fix is making it worse.

FAQ

How much does Lovable cost per month?

You pay $25 a month on Pro for 100 credits and $50 on Business for the same 100. Annual billing runs $250 and $500, shown as $21 and $42 a month, which is two months free. Free costs nothing and gives 5 build credits a day, capped at 30 a month. No paid tier carries a trial period.

What is a Lovable credit?

One unit that pays for five kinds of usage from a single balance: building, chatting, hosting with the built-in backend, model calls your app makes, and managed connector requests. Plan mode charges one credit per message and never touches your code. Build mode charges by the complexity of the request.

Is the Free tier enough to finish a project?

Rarely. It gives 30 build credits a month against the 59 a small internal dashboard needs, so the work takes two calendar months, and it cannot take a custom domain or let you download the code. It is a reasonable place to test an idea. It is not a place to finish one.

What happens when credits run out?

You top up at $0.30 a credit on Pro or $0.60 on Business, or you move up a tier. Auto top-up refills 100 credits when 25 remain, capped at 400 credits of spend a month by default. Unused monthly plan credits expire two months after issue and top-ups after twelve months, and none of it is refundable.

Price the project, not the plan

See what a month of marketing output costs when the research, the content and the reporting run in one place. Plans start at EUR 99 a month, after a 7-day free trial.

Your competitors are already using AllAble. Are you?

The marketers pulling ahead aren't working harder. They're just working with one tool that does everything — that tool is AllAble. Try it yourself!