How to use the Lovable app builder, and what each action costs
The Lovable app builder is a chat panel on the left and a live preview on the right, with Code, History, Security and Publish around them. You prompt in Build mode, answer the questions Lovable asks back, and publish a snapshot to a lovable.app address. The part the tutorials leave out is the price of each move: a Plan message is 1 credit, a Chat message a fraction of one, inline text edits and both security scans are free, and my full builds cost 1.8 to 10.4 credits, read from my own workspace. This guide puts every action on one table, with Lovable's docs re-read on 23 October 2026.
By Marco Kohns · Credits read from the workspace balance before and after each run
1 · Prompt
122free prompts
Paste a structured prompt into the dashboard box. Lovable spots a long one and offers to refine it in Chat first.
2 · Answer
0–4questions
Question cards before any code, when the prompt asks for them. My runs drew up to that many, visual direction included.
3 · Build
1.8–10.4credits
Build mode, billed on the work done, no estimate upfront. Read the cost under each reply in More options.
4 · Publish
Freescans
Publish runs the Quick scan, then deploys a snapshot to lovable.app. Later edits need Publish changes.
Question and credit ranges come from 3 library prompts run between 31 Jul 2026 and 26 Sept 2026.
Pick the right tool
What do you want Lovable to do next?
Use
Build mode
Paste a structured prompt that ends by asking for Lovable's questions. Answer the cards before it writes code.
What it costs
Usage-based, no estimate upfront. Our full builds cost 1.8 to 10.4 credits.
| Situation | Tool | How | Cost |
|---|---|---|---|
| Build a new page or feature | Build mode | Paste a structured prompt that ends by asking for Lovable's questions. Answer the cards before it writes code. | Usage-based, no estimate upfront. Our full builds cost 1.8 to 10.4 credits. |
| A fix missed twice | Plan mode | Describe the symptom and what you tried. Approve the plan, and only then does Lovable switch to Build. | 1 credit per message plus any research. My whole-feature builds cost 9.5 and 10.4. |
| Ask how something works | Chat mode | Ask which file does what, or compare two approaches. Nothing in the code changes. | A fraction of a credit, paid from the daily chat allowance first. Credits used reads Free when the allowance covered it. |
| Change a word or a heading | Inline text edit | Open the preview toolbar, pick text editing, type over the copy and save. | Free for 100 edits per account, each back 24 hours after use. |
| Restyle or move one element | Select element | Press S, click the element (Cmd or Ctrl for several), then describe the change. The selection rides along as context. | Billed as a Build message, priced on the work like any other. |
| Undo the last changes | History | Open History, preview the version, revert. Or edit the message that went wrong and choose Revert and resend. | No refund for the reverted messages, and database data stays as it is now. |
| Check security before launch | Security view | More › Security, run the Deep scan, fix findings from the list. | Scans free, 10 free fixes a day. A security review asked in chat cost us 1.3 credits. |
The order I work in: Chat to settle what I want, Build with a structured prompt, Plan after two misses on the same fix. Lovable's Plan mode docs make the same call on cost: “Chat mode is the cheaper way to ask.”
Measured in my workspace
What does one Lovable prompt cost?
From 1.1 credits for a review asked in chat to 10.4 for a SaaS foundation with sign-in and roles. Each figure is the workspace balance before minus after.
| Action | Credits | Where | Run |
|---|---|---|---|
| Built-in SEO review, asked in chat | 1.1 | Chat, Build mode | 26 Sep 2026, Cadence project |
| Security review, asked in chat | 1.3 | Chat, Build mode | 26 Sep 2026; the free scans live in the Security view |
| Landing page with email sign-ups | 1.8 | Dashboard prompt box | 31 Jul 2026, one round |
| Make the app installable (PWA) | 3.1 | Chat, Build mode | 26 Sep 2026, follow-up on an existing app |
| Six-section SaaS landing page | 4.5 | Dashboard prompt box | 26 Sep 2026, 5 questions first |
| UI states system | 9.5 | Dashboard prompt box | 26 Sep 2026, 2 questions first |
| SaaS foundation with auth and roles | 10.4 | Dashboard prompt box | 26 Sep 2026, 4 questions first |

Lovable prints no estimate before a Build message runs, because, in the words of its Build mode docs, cost “depends on work that is discovered during execution.” The per-message figure sits under every reply: open More options and read Credits used. The pricing math on top of these numbers, plan by plan, is on the Lovable pricing page.
Free or billed
Which Lovable actions are free?
Five are free within limits, five draw credits. Each rule quoted from Lovable's docs, read 23 October 2026.
| Action | Where | Credits | Rule |
|---|---|---|---|
| Quick scan and Deep scan | More › Security | Free | Both scans are free. Deep takes 3 to 13 minutes, at most 15. docs |
| Comment on the preview | Preview toolbar | Free | Adding comments is free; sending a thread to Lovable is billed as build. docs |
| Inline text edit | Preview toolbar | Free | 100 free per account, each back 24 hours after use; then billed as build. docs |
| Try to fix, or a fix from Security | Chat, Security view | Free | 10 free fixes per account, each back after 24 hours; then billed as build. docs |
| Edit code yourself | Code tab | Free | No credits. Editing needs a paid plan; Free is read-only. docs |
| Chat mode message | Mode picker | Billed | Priced on the work, typically a fraction of a credit; the daily chat allowance pays first. docs |
| Plan mode message | Mode picker | Billed | 1 credit, plus any subagent research. Each plan revision is a new message. docs |
| Select element or Draw annotation | Preview toolbar | Billed | Billed as a Build message. docs |
| Publish or edit metadata from the chat | Chat | Billed | Billed as build. Publishing from the Publish button is free, per Lovable's product guide. docs |
| Revert to an earlier version | History | Billed | Refunds nothing: the reverted messages still count. Restores code, never database data. docs |
Where to read the cost of one message
- Open More options under Lovable's reply.
- Credits used shows the charge so far while it runs, and the final cost when it ends.
- In Chat mode it reads Free when the daily chat allowance covered the message.
- More › Usage lists the project's cost credit by credit, split into Build and Run.
Docs: project chat, editor, read 23 October 2026.
When one message runs long
- A Build message can work for up to 10 hours, and a long one costs more.
- A credit check-in pauses it at 20 credits by default and asks before going on (beta).
- The stop button ends it at any time; what it already charged stays charged.
- Typing /goal makes Lovable work until a goal is met, which the docs warn can use a large number of credits.
Docs: Build mode, read 23 October 2026.
The editor
What are the parts of the Lovable editor?
Chat on the left, the preview on the right, four tabs and two controls around them. The backend (database, users, storage) lives under More.
| Part | Where | What it does | Cost |
|---|---|---|---|
| Chat panel | Left | Where you prompt, answer Lovable's question cards and follow progress. The mode picker and the plus menu sit under the input. | Each message priced by mode |
| Preview | Right, home tab | The app running live, with Desktop, Tablet and Mobile views. | Free to use |
| Preview toolbar | Bottom of the preview | Select elements, edit text inline, draw an annotation, comment. | Text edits and comments free (limits apply); Select and Draw billed |
| Files | Toolbar tab | Documents Lovable generates and files you upload. | Free to open |
| Code | Toolbar tab | Browse, search and download the code; edit it on paid plans. Each save is a version. | Manual edits free |
| More | Toolbar tab | Analytics, Database, Users, Storage, AI, Payments, Emails, Connectors, Security, Usage. | Running the app draws Run credits |
| History | Top bar toggle | Open any past version, revert to it, edit a past message and resend. | Revert refunds nothing |
| Publish | Top right | Deploys a snapshot to a lovable.app address; later edits need Publish changes. Runs the Quick scan first. | Custom domain needs a paid plan |


Two rules decide most surprises. Reverting in History restores code only, so rows a message wrote to the database stay, and the history docs are plain that reverted messages are never refunded. And the Code tab is read-only on Free: manual edits cost no credits, but only on a paid plan, and where the database and its policies live is a separate guide.
One project, start to finish
What does a first Lovable project cost, prompt to installable app?
20.4 credits over 5 steps on 26 September 2026, one project called Cadence: the foundation, two reviews, a phone install and a landing page.
- 12:26FoundationAsked 4 questions, enabled Cloud, self-tested sign-up and invites, roles checked in the database10.4
- 14:29SEO review5 of 6 checks passing; the one failure was the sitemap, deferred until publish1.1
- 18:06SecurityThe chat ran its own checks (clean) and said the Deep scan must start from the Security tab; run there, free, 3 minutes, no findings; the dependency panel listed 2 known issues on one package1.3
- 18:36InstallableManifest, icons, service worker gated to the published app, Add to Home Screen button; verified in the preview3.1
- 19:19Landing pageFive pre-build questions, six minutes to a preview; server-rendered, FAQ schema, one CTA type; 394 words4.5
The full ledger with every prompt verbatim is on the MVP guide. The two steps I would change: the security review asked in chat cost credits for a check the Security view runs free, and the landing page went into its own project, so the product now has two projects and two addresses.


Habits from the runs
How do you spend fewer credits in Lovable?
Each line comes from a run on this site or a rule in the docs, and the cheap move is usually the one the editor already offers.
Do
The cheap path
- End every build prompt with a line asking for Lovable's questions first: my SaaS run asked 4 before writing code.
- Fix copy with inline text edits, 100 of them free per account.
- Run Quick and Deep scans from More › Security, both free.
- Ask in Chat before you commit to a plan; the daily allowance pays first.
- Name the states, roles and build order in the prompt, as the library prompts do.
- Read Credits used after each message and keep a ledger per project.
Avoid
The expensive path
- Re-prompting a whole feature after two misses. Switch to Plan for 1 credit plus research.
- Asking the chat for a security review: billed, while the scans are free.
- Reverting to get credits back. Nothing is refunded.
- Reverting after a message changed data, expecting the rows to roll back too.
- Publishing or editing metadata through the chat, which bills as build.
- Describing an element in a paragraph when Select element can point at it.
How to write the prompt itself, section by section, is in the Lovable prompting guide; eight weak prompts and their rewrites are in the prompt examples.
Start with a prompt that asks first
122 free library prompts, each ending in the line that asks Lovable for its questions before it builds.
Disclosure: if you create a Lovable account through links on this page, we may earn a commission. It costs you nothing and does not change what we recommend. Create a free Lovable account
Frequently asked
How do I start using Lovable?
Sign up, paste a prompt into the box on the dashboard, answer the questions Lovable asks, and watch the preview build. Free gives 5 build credits a day, enough for one structured landing page prompt: ours cost 1.8. Lovable's own prompting guide opens with the same advice we give: make Lovable ask clarifying questions before it builds.
Is Lovable's app builder free?
The editor, the preview, both security scans, comments and the first 100 inline text edits are free on every plan. Building spends credits, and editing code by hand needs a paid plan, because the Free code editor is read-only. Plan prices and the credit ladder are on our pricing page.
What is the difference between Build, Plan and Chat mode in Lovable?
Build changes your project and is the default. Plan investigates and writes a plan you approve before any code changes, at 1 credit per message plus research. Chat discusses the open app without editing and costs a fraction of a credit from the daily chat allowance. Source: Lovable's project chat docs, read 23 October 2026.
How do I see what one Lovable message cost?
Open the More options menu under Lovable's reply. It shows Worked for and Credits used, live while the message runs and final when it ends. That one number is how every figure on this page was checked, and it lets you price your own prompts the same way.
Does Lovable stop an expensive message?
Yes, with a credit check-in: when one message crosses your check-in level, 20 credits by default, Lovable pauses, shows the cost so far and asks whether to continue. The feature is in beta and the level is set in account settings. The stop button ends a message at any time.
Can Lovable edit an existing GitHub repository?
Not directly. Git sync always creates a new repository and syncs one branch at a time; to start from existing code, Lovable's GitHub docs (read 23 October 2026) say to download the repository as a ZIP, import it as a project, then connect Git sync to that new project. Tools that open an existing repo are on our Lovable vs Bolt and Lovable vs Cursor pages.
Where do I put instructions Lovable should always follow?
In project knowledge (up to 10,000 characters, wins on conflict) or workspace knowledge (also 10,000) for rules across projects. A root AGENTS.md file in the code is always read too. Skills hold instructions that only matter for some tasks and load on demand.
What we did not publish
- Measured figures for a single Plan message, a Chat message, a Select edit or a publish. The table gives the docs rule for those until I run each one.
- Screenshots of the More panel after its October redesign; ours is from 26 September 2026.
- Mobile app editing. The phone app has its own guide.
- Runs on Bolt, Replit or v0. Only Lovable runs are mine.
Sources
- Lovable docs: Editor
- Lovable docs: Project chat
- Lovable docs: Build mode
- Lovable docs: Plan mode
- Lovable docs: Preview toolbar
- Lovable docs: Code editor
- Lovable docs: Version history
- Lovable docs: Publish
- Lovable docs: Security view
- Lovable docs: Knowledge
- Lovable docs: GitHub
- Lovable docs: Prompting best practices
- My runs: 31 Jul 2026, 26 Sept 2026, 26 Sept 2026, and the Cadence ledger on 26 September 2026
Docs read 23 October 2026.
Written by

Marco Kohns
Founder of ProtoBites - Venture Growth Studio
Growth PM at a Silicon Valley scale-up (a16z and General Catalyst backed), ex-Techstars where he consulted 13 early-stage startups, Reforge-trained. Every prompt on this site comes out of shipping ProtoBites' own portfolio products.
Related reading
- every Lovable guide, by build stage · Choose a tool, plan, prompt, build safely, launch.
- app builder prompts · Thirty whole apps to paste into the chat panel.
- what to type into the chat panel · The prompting guide.
- Skills, goals and Plan mode · The controls beyond Build.
- Lovable pricing and credits · Plans read today and what one prompt costs in credits.
- three prompts and what they built · The runs the screenshots come from.
- the SaaS prompt built in these screenshots · Cadence, workspace and invites.
- Lovable vs Cursor · When you want an editor instead.
- build kits · Schema files to open in the code view.
- a dashboard schema to open in the code view · The admin dashboard kit.
- Lovable SEO and the built-in review · SSR, pre-rendering and our migration data.
- Lovable mobile app · The phone app's limits from the docs, and PWA versus Capacitor.
- Lovable vs v0 · How v0's Design Mode compares with Lovable's visual edits.
- what is vibe coding · The term behind the builder, from Karpathy to Collins.
- what is an AI app builder · What an AI app builder is, from the outside.
- an honest Lovable review · Measured builds, ratings and who should skip it.