🧠 Introducing OM2: Your enterprise just started thinking.Read more

On this page

Put Coworker to work on your stack.

Connect Salesforce, Slack, Jira and run your first agent in minutes.

Book a demo
Blog

Enterprise AI

Lovable Pricing in 2026: What Credits Actually Cost

Lovable is $25/mo Pro and $50/mo Business, both at 100 credits. Here is what burns credits fastest, what rolls over, and what a real shipping month costs.

Dhruv Kapadia27 min read

What Lovable costs right now

Lovable pricing is published openly, which already puts it ahead of most tools in this category. As of September 18, 2026, Lovable's pricing page lists four plans:

PlanEntry priceCredits includedNotable limits
Free$0/month5 build credits per day, capped at 30/monthLovable badge on your app, no credit rollover, no top-ups
Pro$25/month100 monthly creditsCustom domains, badge removal, credit rollover, top-ups
Business$50/month100 monthly creditsEverything in Pro plus SSO, security center, design templates
EnterpriseCustom platform feeVolume-basedSCIM, audit logs, data residency, named CSM

The catch is that $25 and $50 are the floor of a selector, not a plan price. Lovable's own subscription docs list eleven credit tiers on each paid plan. Pro runs from 100 credits at $25 to 10,000 credits at $2,250 per month. Business runs from 100 credits at $50 to 10,000 at $4,300. Annual billing is two months free, so Pro at 100 credits is $250 a year, or about $21 a month.

Business costs exactly twice Pro for the same credit count. Lovable's pricing FAQ explains why that is less punitive than it looks: the dollar cost of hosting and running your app is the same on both plans, so because Business credits are worth more, a Business workspace burns fewer credits for identical runtime.

Every Lovable pricing tier, side by side

The pricing page shows one entry price per plan and a selector. Lovable's subscription docs publish the whole ladder, which is the table most buyers never see. These are the eleven Pro tiers as of September 18, 2026.

Monthly creditsPro, billed monthlyPro, billed annuallyBusiness, billed monthly
100$25$250 ($21/mo)$50
200$50$500 ($42/mo)$100
400$100$1,000 ($84/mo)$200
800$200$2,000 ($167/mo)$400
1,200$294$2,940 ($245/mo)$588
2,000$480$4,800 ($400/mo)$960
3,000$705$7,050 ($588/mo)$1,410
4,000$920$9,200 ($767/mo)$1,840
5,000$1,125$11,250 ($938/mo)$2,250
7,500$1,688$16,880 ($1,407/mo)$3,300
10,000$2,250$22,500 ($1,875/mo)$4,300

Two things fall out of that table that the marketing does not say.

The volume discount is small. On Pro, a credit costs $0.25 at 100, 200, 400 and 800 credits. It only starts falling at 1,200, and bottoms out at $0.225 at the top tier. Buying 100 times more credits saves you 10% per credit. If you are waiting for scale to make the unit cost sensible, it does not arrive.

Annual billing beats volume. Pro at 100 credits annually is $250 for 1,200 credits a year, which is about $0.21 per credit, cheaper than the largest monthly tier. If your usage is steady, the billing cycle is a bigger lever than the tier.

Business runs at exactly double Pro per tier until the top, where $4,300 against $2,250 is a little under double. Same credit count, roughly twice the price, and the difference buys governance rather than capacity.

How Lovable's credit system actually works

This is the part buyers get wrong, so it is worth being precise. Lovable unified its billing in 2026 into a single credit balance that pays for three different things: building your app, hosting and running it on Lovable Cloud, and any AI features your deployed app calls at runtime.

There are two kinds of credits and they behave differently.

Usage-specific grants are included allowances tied to one purpose. Every plan gets 5 build credits per day that reset at 00:00 UTC, 20 Cloud credits per month, and 4 AI credits per month. Lovable spends these first. Per the credits documentation, none of them roll over. Free plans cap the daily build grant at 30 per calendar month, which works out to six days of building; Pro and Business have no monthly cap on the daily grant.

General credits are the ones you pay for. They arrive at the start of each billing period and cover anything the grants did not.

Build cost depends on which mode you are in. Plan mode is a flat 1 credit per message, plus the cost of any subagent research Lovable runs while planning. Build mode is usage-based, priced on the complexity of the request and how much work got done.

What burns credits fastest

Lovable publishes four worked examples on its pricing page, and they are the most useful numbers in this whole post:

PromptWork doneCredits
"Make the button gray"Updates the button styles0.50
"Remove the footer"Removes the footer component0.90
"Add authentication with sign up and login"Adds auth pages, logic, routes1.20
"Build me a landing page, use images"Landing page, 3 generated images, theme, 5 sections1.70

So a meaningful build message averages roughly 1 credit. That sounds cheap and it is, right up until debugging starts. You are charged for work completed, not for work that succeeded, so a failed fix costs the same as a working one. A loop where the model breaks something, you ask it to fix that, and it breaks something adjacent is the single fastest way to drain a balance.

The second drain is runtime. Once an app is live, Cloud and in-app AI usage draw from the same balance your builds do. The 20-credit monthly Cloud grant covers a quiet app. It does not cover a busy one, and Lovable does not publish a rate card for Cloud consumption, so this is the one part of the bill you genuinely cannot forecast before you ship.

How to make your Lovable credits last

This is the part worth reading twice, because most of the levers are free and most people find them after the first surprising invoice. Every item below comes from Lovable's own documentation.

Use the 10 free fixes before you spend a credit on a bug. Per Lovable's chat documentation, your account includes 10 free fixes shared across the Try to fix button and fixes suggested by security scans, counted across every workspace and project rather than per project, and each one becomes available again 24 hours after you use it. After the 10 are spent, further fixes bill as ordinary build usage. Security scans themselves are always free and do not count against the 10.

Revert instead of repairing. Lovable's debugging guide says it plainly: if a series of fixes has tangled the code, rewinding is usually faster than untangling. Restore a working state from version history, or edit an earlier message and choose Revert and resend. Then tell Lovable what you rolled back so it does not rebuild from a stale picture. A revert costs nothing. The fourth attempt at the same fix does not.

Plan in Plan mode, build in Build mode. A Plan mode message is a flat 1 credit whether it ends in a plan, a clarifying question or an answer, and skipping the plan adds no charge. The catch is subagent research: Lovable's Plan mode docs say broad, open-ended requests are the ones likely to trigger it, and that research is usage-based. Point planning requests at specific files or features and the flat credit is what you pay.

Make it ask you questions first. Lovable's prompting guide recommends ending a prompt with a line asking Lovable to ask whatever it needs to understand the feature. Clarifying before generating is cheaper than regenerating after, because Build mode bills for work completed rather than work that turned out to be right.

Prompt by component, not by page. The same guide is blunt about this: a full-page prompt returns noise, a section-based prompt returns signal. Build the hero, review it, then the feature grid. When something is wrong you repair one block rather than regenerating the page around it.

Write real copy, not placeholder text. Lorem ipsum gives the model nothing to lay out against, so you pay twice: once for the generic version and again for the version built around your actual headline.

Turn on credit check-ins and leave them on. Lovable pauses a long-running Build message when it crosses your check-in level, 20 credits by default, and shows what the message has cost so far, with Continue and Finish up as the options. A single message can run for up to 10 hours, so this is the difference between noticing a runaway job and reading about it on the invoice. The level is configurable between 20 and 100,000 credits in account settings.

Be careful with `/goal`. A message set as a goal is billed on the work done like any other, but Lovable keeps working until the goal is met, and its own documentation says the same work costs noticeably more this way.

Cap the blast radius in a shared workspace. Credits are pooled across everyone you invite, so workspace owners and admins can set a default monthly credit limit per member plus per-member overrides. Auto top-up has its own monthly spend limit, which defaults to 400 credits and resets on the 1st of each calendar month rather than on your billing cycle. If nobody sets these, one person's bad week is the whole workspace's balance.

Buy the tier, not the top-up. Subscription credits at the Pro base tier work out to $0.25 each. Top-ups are $15 per 50 credits on Pro and $30 per 50 on Business, which is $0.30 and $0.60 per credit. Lovable's own documentation says that if you consistently need more credits each month, upgrading is usually better value, and the arithmetic agrees.

One thing you cannot optimise away: stopped Build requests are charged for the work completed up to the stop. Hitting the stop button limits the damage, it does not undo it.

Coworker

See what your AI stack really costs

Compare model and platform costs, then run it all in one place.

Open the free LLM cost calculator
Book a demo

Do unused Lovable credits roll over?

Partly, and the distinction matters. Unused monthly plan credits roll over at the end of each billing cycle for as long as your subscription stays active. They still expire on their own schedule: two months after issue on monthly plans, one month after the annual period ends on annual plans. Top-up credits last 12 months.

Daily build credits do not roll over. Five a day, use them or lose them. Over a 22-day working month that is 110 credits of free building you forfeit by not showing up, which is more than a full month of the base Pro tier.

Top-ups are available on Pro and Business only, at $15 per 50 credits on Pro and $30 per 50 on Business. Do the arithmetic against the subscription rate and the premium is clear: Pro subscription credits at the base tier cost $0.25 each ($25 for 100), while Pro top-ups cost $0.30 each. That is 20% more for the same credit. If you top up regularly, you are paying a surcharge to avoid moving up a tier.

What Lovable pricing looks like for someone actually shipping

Here is a model with the assumptions stated, using Lovable's own published per-prompt numbers.

Assume one person building actively: 20 build messages on a working day, at the ~1.1 credit average implied by Lovable's four examples. That is about 22 credits a day, or 484 credits across 22 working days. The daily grant covers 5 a day, so 110 of those are free, leaving roughly 374 credits to come out of the plan balance. Add some Cloud draw past the 20-credit grant once the app is live.

The nearest Pro tier that covers that is 400 credits at $100 a month, not $25. Covering the same usage with top-ups instead, which sell in 50-credit blocks, means 400 credits at $120.

The light case is very different. Someone making 5 build messages a day sits almost entirely inside the free daily grant and spends close to nothing from the plan balance. For that person the $25 Pro tier is real, and they are mostly paying for custom domains and badge removal.

The spread between those two users is the whole story of Lovable pricing. Same tool, same sticker, four times the bill.

Four worked scenarios, with the assumptions stated

None of this is survey data. It is arithmetic on Lovable's published per-prompt examples, so swap in your own message volume and the shape holds. I am using roughly 1.1 credits per substantive build message, the average of Lovable's four published examples, and 22 working days a month.

WhoBuild messages per working dayCredits needed from the planNearest Pro tier
Weekend tinkerer5, mostly inside the daily grantclose to 0Free, or $25 for a custom domain
Evening side project10, five of them covered by the grantabout 120$25 (100 credits) plus a top-up, or $50
Solo builder, full time20about 374$100 (400 credits)
Two people on one product20 eachabout 750$200 (800 credits)

The daily grant is doing most of the work in the top two rows. Five build credits a day on every plan, reset at 00:00 UTC, is 110 credits over 22 working days, which is more than the entire base Pro tier. Someone who builds in short daily sessions can run a real project for the price of the custom domain and the badge removal.

The bottom two rows are where the sticker price stops describing anything. A full-time solo builder needs about four times the entry tier. Two people sharing a workspace need eight times it, and that is before any Cloud draw from a live app.

The number that should drive your tier choice is messages per day, not headcount and not the size of your idea. Lovable prices by work performed, so a careful builder on a big product can cost less than a scattershot one on a small product.

How Lovable pricing compares to Bolt, Replit, v0 and Cursor

These tools all meter something, but they do not meter the same thing, which makes the entry prices close to useless as a comparison. Here is what the money actually buys, read live on September 18, 2026 from each vendor's own pricing or billing documentation.

ToolEntry paid priceWhat that buysFree planUnused allowance
Lovable$25/mo Pro100 credits, about 90 substantive build messages$0, 5 build credits/day capped at 30/moRolls over, expires 2 months after issue
Bolt.new$25/mo Pro10M tokens, spent mostly on syncing project files$0, 1M tokens/mo with a 300K daily capRolls over, valid 2 months, oldest spent first
Replit$20/mo Core$20 toward paid model usage, plus Free Mode allowanceStarter, daily capPro tiers roll over for two months
v0$30/user/mo Plus$30 of credits per user, billed against published token rates$0, $5 of monthly credits, 7 messages/dayRolls over, expires after 65 days
Cursor$20/mo ProIncluded usage across two model pools, not a credit balanceHobby, limited agent requestsPricing docs describe on-demand overage, not rollover

Four observations that matter more than the prices.

Lovable and Bolt are the same $25 for very different units. Lovable charges for work completed and tells you the cost of a message before you send a similar one. Bolt charges for tokens processed, and its own FAQ says most of that is syncing your project files to the model, so the same prompt gets more expensive as your codebase grows. Lovable's meter moves with what you ask for. Bolt's moves with how big your project already is.

v0 is the only one that publishes a rate card. Its pricing page lists per-million-token input and output rates for each model, from $0.20 and $1.20 on v0 Mini to $10 and $50 on v0 Max Fast. Credits are denominated in dollars, so $30 of credits means $30 of model usage at those rates. If you want to forecast before you build, that is the most forecastable model in the category.

Cursor is the odd one out, and Cursor says so. Its pricing documentation tells you what to expect rather than what you are buying: daily agent users typically land at $60 to $100 a month of total usage, and power users running multiple agents often exceed $200. Pro is $20, Pro Plus is $60, Ultra is $200, and Teams seats are $40 or $120 per user. It is a code editor with agents rather than an app builder, so it belongs on this list as a budget comparison, not a feature one.

Only Replit and Lovable bill hosting from the same balance as building. On both, a live app with real traffic quietly draws down the balance you were using to build. On Bolt, hosting is bundled with the subscription up to published bandwidth and request limits. That single design choice is why Lovable bills surprise people who ship successfully and Bolt bills surprise people who build for a long time.

What users say about the cost

Credit burn is the consistent complaint, and it is worth taking seriously rather than dismissing. In a review published July 14, 2026, Laura Wendel described a multi-user portal that "ate through 40 credits in an afternoon" fixing issues the AI had introduced. Worth noting that Zite sells a competing product, so read it with that in mind, but the specific number is consistent with what Lovable's own per-prompt pricing implies once a fix loop starts.

The pattern repeats in G2's Lovable reviews. A verified DevOps engineer reviewer on August 31, 2026 wrote that "credit consumption can also become difficult to predict during longer development cycles, especially when debugging or iterating on complex features," and asked for better visibility into what is consuming credits. The complaint is almost never about the $25. It is about the gap between the $25 and the total once an app is live.

Two of Lovable's older documentation pages still describe the pre-unification model, including a claim that top-ups are unavailable. They are out of date. When two Lovable sources disagree, the live pricing page wins.

What changed in Lovable's pricing, and why it matters now

Lovable moved to a single credit balance on June 13, 2026. Before that, a workspace carried two balances: build credits for developing the app, and a separate dollar-denominated Cloud and AI balance for everything that happened once the app was live. Any purchased Cloud and AI dollars were converted into credits at the plan's credit price and added to the balance.

Lovable's announcement is specific that plan prices, the cost of a credit, the daily free credits and the expiry rules did not change. What changed is which pool pays for what, and the answer is now one pool for everything.

That reframes the buying decision. Under the old model you could treat building and running as separate budgets, and a runaway hosting bill could not stop you building. Under the new model, a busy live app and a heavy build week compete for the same credits. The 20-credit monthly Cloud grant and the 4-credit monthly AI grant exist to keep small apps off the main balance, and Lovable's documentation calls both a temporary offering subject to change.

One more piece of history worth knowing before you plan around a number: the Cloud and AI usage history in your dashboard starts on the date your workspace moved to the unified balance, June 1, 2026 at the earliest, and credit history recording began on March 8, 2026. If you are looking at a usage chart and reasoning about a trend, check that the chart is old enough to have one.

The vendor's own sources also disagree with each other in one small place. The pricing page lists "Build me a landing page, use images" at 1.70 credits. The credits documentation lists the same landing page example at 2.00 and labels all four numbers illustrative. Both are Lovable pages. The live pricing page is the one to plan against, and the gap between the two is a fair estimate of how much these numbers move.

I built a prototype. What does it cost to actually run it?

This is the question the sticker price answers least well, because on Lovable the app you shipped keeps spending after you stop building it.

Running costs are billed as run credits, and Lovable's project usage documentation breaks them into six categories: database server, database storage, network, storage, compute and realtime, plus AI gateway usage when features inside your published app call a model. The database instance size and how long it stays active are the drivers people underestimate, because they are charged per hour of running rather than per visitor.

Lovable does not publish a rate card for any of this. You get a 20-credit monthly Cloud grant and a 4-credit monthly AI grant, and after that it comes out of the same balance you build with. So the honest answer to "what will hosting cost" is that you cannot know before you ship, and you should watch More, Cloud, Usage in the first week rather than model it in advance.

What happens if the balance hits zero is more nuanced than most write-ups suggest. Per the same documentation, your published site stays live. AI features inside the app stop working, and apps that depend on the built-in backend can pause until credits are added. A static marketing page survives an empty balance. A logged-in product does not.

The checklist Lovable publishes for bringing a project's usage down, in its own order of impact:

  1. Pause the project's Cloud backend when it is not in use, which mainly helps for projects with scheduled jobs or other background activity, since Lovable already pauses inactive Cloud projects on its own.
  2. Right-size the database instance, because a larger instance costs more credits per hour that your app runs.
  3. Fix slow database queries so each request does less work. You can ask Lovable in the project chat to find and optimise the slowest ones.
  4. Delete unused files from storage. Stored files keep costing credits even while a project is paused.
  5. Review scheduled jobs. Frequent or long-running ones add up quietly.
  6. Remove Lovable Cloud from projects that never needed a backend. This permanently deletes that project's Cloud data, so export first.

Two governance details matter if you are running this inside a company. Per-member monthly credit limits apply to build credits only, so a member limit does not cap what a live app spends. And downloading your code or syncing to GitHub, GitLab or Bitbucket is how you keep the option to host somewhere else: Git sync is available on every plan including Free, while downloading the code requires a paid plan.

Who Lovable is genuinely wrong for

Worth saying honestly, because the tool is good at what it is good at and the mismatches are predictable.

Anyone who needs a fixed monthly number. Lovable bills for work completed, including work that failed, and does not publish a rate card for hosting. If your budget process needs a figure you can defend in advance, a seat-priced tool will be easier to sign off than a meter with an unpublished component.

Teams where one person debugs by re-prompting. The meter does not care whether the fix worked. A person who responds to a broken build by asking again, three or four times, costs several times what a person who reverts and re-plans costs, on the same product. If you cannot change that habit, the tool will keep charging for it.

Projects that are mostly maintenance. The daily grant of 5 build credits is generous for someone making small, steady changes and irrelevant to someone doing a large migration. If your work is long, rare, heavy pushes rather than daily edits, most of that free allowance expires unused at 00:00 UTC every night.

Anyone who needs the app to survive an empty balance. If an unpaid invoice or a spending cap can pause your product's backend, that is a business risk rather than a pricing preference. Export the code and host it yourself before it matters, not after.

Buyers who want per-seat governance rather than shared credits. Lovable does not charge per seat. Every plan supports unlimited workspace members who share one pool, which is excellent for a class or a small team and awkward for an organisation that budgets per head. Per-member limits exist, but someone has to set them.

Where Lovable pricing fits next to the rest of your AI stack

If you are pricing Lovable, you are probably also pricing an AI coding assistant, and the two models are not comparable. Lovable meters work performed. Most assistants meter seats. Our Windsurf pricing breakdown covers the seat-based side of that comparison, and the enterprise AI pricing comparison puts both models next to each other.

Coworker solves a different problem from Lovable, and it is worth being clear about that rather than pretending they compete. Lovable builds the app. Coworker MCP gives AI tools like Claude Code, Cursor, ChatGPT, and Windsurf unified context across the systems your company already runs, through one governed connection to 50+ apps. If the thing slowing your team down is that the AI does not know what is in your Jira, your Slack, or your CRM, that is the gap Coworker fills, and it is orthogonal to whichever app builder you pick. Coworker is $29.99 per user for Pro, $149.99 for Max, and custom for Enterprise, priced per seat rather than per credit.

If the protocol itself is new to you, start with what MCP is, then how to choose MCP servers for evaluating options.

Book a demo to see how Coworker MCP works against your own systems.

Frequently asked questions

How much does Lovable cost?

Lovable's Free plan is $0 with 5 build credits a day capped at 30 a month. Pro starts at $25 a month and Business at $50 a month, both at 100 credits. Both paid plans scale through eleven credit tiers, up to $2,250 a month on Pro and $4,300 on Business at 10,000 credits.

Do Lovable credits roll over?

Unused monthly plan credits roll over while your subscription is active, expiring two months after issue on monthly plans. The 5 daily build credits do not roll over and reset at 00:00 UTC. Top-up credits last 12 months from purchase.

What uses the most credits in Lovable?

Build mode work scales with complexity, and you are charged for work completed whether or not it fixed the problem, so debugging loops are the biggest drain. After that it is runtime: once your app is live, Cloud and in-app AI usage draw from the same balance, and Lovable does not publish a rate card for that consumption.

Is Lovable Business worth double the price of Pro?

Business costs twice Pro for the same credit count, and adds SSO, a security center, personal projects, internal publish, and design templates. Lovable's own FAQ notes that the dollar cost of running an app is identical on both plans, so a Business workspace consumes fewer credits for the same runtime. Whether that offsets the price depends on how much of your usage is runtime versus building.

What does a realistic month on Lovable cost?

Using Lovable's published per-prompt examples, a solo builder making about 20 build messages a working day needs roughly 400 credits a month after the free daily grant, which is the $100 Pro tier. Someone making 5 messages a day stays inside the free daily grant and the $25 tier is genuinely enough. State your own message volume before picking a tier.

Is there a cheaper way to buy Lovable credits?

Yes. Subscription credits cost less per credit than top-ups. At the Pro base tier subscription credits work out to $0.25 each while top-ups are $0.30, a 20% premium. If you top up most months, moving up a credit tier is cheaper than the surcharge.

Is the Lovable free plan enough to build a real app?

For a prototype, often yes. The Free plan gives 5 build credits a day capped at 30 per calendar month, which works out to six days of building per month, plus a 20-credit monthly Cloud grant and 4 AI credits. The limits that usually push people to Pro are not the credits, they are the missing custom domain, the Lovable badge on the published app and the inability to download your code.

How much does it cost to run a Lovable app after I publish it?

Lovable does not publish a rate card for it. Hosting and the built-in backend are billed as run credits across database, storage, network, compute and realtime usage, with a 20-credit monthly Cloud grant included on Free, Pro and Business. Check the Usage view on the project in the first week after launch rather than trying to forecast it.

Does Lovable charge per seat?

No. Every plan supports unlimited workspace members and prices by credits, so inviting people does not change the subscription cost, only how fast the shared pool empties. Workspace owners and admins can set a default monthly credit limit per member plus per-member overrides, and those limits apply to build credits only.

Does Lovable offer a student discount?

Yes. Students and teachers with a verified academic email address get 50% off for up to 12 months on the Pro plan at 100 credits a month, billed monthly. The discount does not apply to higher credit tiers, annual billing or top-ups, is tied to the account that verified, and is not offered again after you cancel.

How do I stop a debugging loop from draining my Lovable credits?

Use the 10 free fixes first, and note that each one returns 24 hours after it is used. If the same error survives two attempts, revert to a working version rather than prompting again, then tell Lovable what you rolled back. Turn credit check-ins on so a long message pauses at 20 credits instead of running for hours. Build mode charges for work completed, not for work that succeeded, so the loop is the expensive part, not the bug.

Is annual billing worth it on Lovable?

It is the largest discount available. Pro at 100 credits is $250 a year against $300 billed monthly, about $0.21 per credit, which beats even the 10,000-credit monthly tier at $0.225. The trade is expiry: credits on an annual plan expire one month after the annual period ends, versus two months from issue on monthly plans.

Lovable or Bolt.new: which is cheaper?

They both start at $25 a month and they meter different things, so it depends on your project's shape. Lovable charges for the work each message performs, so cost tracks what you ask for. Bolt charges for tokens, most of which go to syncing your project files, so cost tracks how large your codebase already is. A long-lived, growing project tends to get more expensive on Bolt. A project with heavy live traffic tends to get more expensive on Lovable, because hosting draws from the same credit balance.

Ready to get started?

Put Coworker to work inside your actual stack

Connect Salesforce, Slack, Jira, whatever you use, and run your first agent in minutes.