How Much Does an AI-Built Website Cost? Real Numbers
An AI-built website is priced per action rather than per page. On Zugo a single-page build costs 6 credits, an edit costs 3, and a multi-page platform costs 12 for the first three pages plus 3 for each page after that. Pro is $25 a month for 200 credits. Free comes with 5.
That is the whole price list, and it is short on purpose. The interesting question is not the sticker price but the total: how many edits a real site takes, what the connectors add, and where this way of building stops being the cheap option.
What does an AI-built website actually cost?
Two numbers decide your bill: what you build and how many times you change it. Nothing is charged per token, per seat, or per visitor, which is why you can estimate a project before starting it rather than after.
| Action | Credits |
|---|---|
| Single-page site, app, or game | 6 |
| Multi-page platform, first three pages | 12 |
| Each page beyond the third | 3 |
| Any edit | 3 |
| Hi-Fi build | 12 |
| Hi-Fi edit | 6 |
Hi-Fi is the higher-effort mode, and it costs exactly double. It is worth it on the pages a customer judges you by, and wasteful on an internal admin screen nobody but you will see.
A five-page platform is therefore 12 plus two extra pages at 3 apiece, which is 18 credits before you have changed anything. Add the ten edits a real site needs and you are at 48. That arithmetic is the honest version of "how much does a website cost", and you can do it yourself in advance.
How do credits turn into dollars?
Divide the plan by its allowance. Pro is $25 for 200 credits, so a credit works out at 25 divided by 200, which is 12.5 cents. Every action price above converts straight from that.
At that rate a single-page build is 75 cents, an edit is a little under 40 cents, and a three-page platform is a dollar and a half. The five-page platform with ten edits from the previous section comes to six dollars. Those numbers are small enough that the real budget question is subscription tier, not action count.
Pro carries 200 credits a month, which is roughly 16 full platforms or 33 quick builds if you spend the whole allowance on generation and never edit. Nobody does that, so treat it as a mixed budget: a handful of projects and the edits they need. Business is $99 a month for teams and heavier use.
Free comes with 5 credits and no card. That is enough to see the interface and a real result on your own idea rather than a canned demo, and not enough to finish a project, which is what a free tier honestly is. Our full pricing breakdown goes through the tiers in more detail.
What does a real project cost from start to publish?
Three worked examples, all using only the prices above, all assuming a normal amount of changing your mind.
A one-page landing for a service. One build at 6, six edits at 3 for copy and layout, so 24 credits. Three dollars at Pro rates, published the same evening.
A four-page company site. A platform build at 12 covers the first three pages, the fourth adds 3, and twelve edits across all of them add 36. That is 51 credits, or a bit over six dollars.
A small app with logins. Same 12 to start, a few extra pages, and considerably more editing, because auth rules and data models are exactly where the first draft misunderstands you. Budget twenty to thirty edits and you are still under twenty dollars of allowance.
The pattern in all three: generation is the cheap part and editing is the real cost. That is also true of hiring a person, it is just easier to see here because each change has a number on it.
What keeps costing money after the site is live?
The site itself is published and served, so there is no per-visitor bill from the builder. What continues costing money is change and the services you connect.
Connectors are your accounts, not ours, and they bill you directly. Supabase covers the database, login, and file storage. Stripe takes its cut of payments. Resend sends email. Vercel hosts if you deploy to your own account. Google Analytics is free. Most small sites sit inside the free tiers of all of these for a long time, but the bills are separate and they are yours.
A custom domain is a registrar fee, typically annual, and again not ours. Published projects get a your-project.zugo.run address immediately, so the domain is a choice about branding rather than a requirement for going live. If you want one, connect it before you start sharing links.
Then there is the honest ongoing item: edits. A site you touch twice a year costs almost nothing to keep. A site you restructure monthly is a subscription decision, and Pro exists precisely for that shape of use.
How does this compare with the other ways to get a site?
Every option here is genuinely better than the others at something, and pretending otherwise would not help you choose.
| Route | Rough cost | Where it wins | Where it does not |
|---|---|---|---|
| AI builder | Tens of dollars, same day | Speed, cheap iteration, you keep the code | Complex logic needs many rounds |
| Template site builder | Monthly fee, a weekend | Predictable output, mature editors | You work inside the template's limits |
| Freelancer | Hundreds to low thousands | Judgment, custom work, someone accountable | Slower to start, revisions cost real money |
| Agency | Thousands and up | Strategy, brand, ongoing support, scale | Overkill for a first version |
A freelancer or an agency brings something no builder does: a person who tells you the idea is wrong before you pay to build it. That is worth real money on a project with a real budget behind it, and it is the main reason to skip this route entirely.
Where the AI builder wins is the first version. Getting something published and in front of actual visitors this week, for the price of a lunch, is a different kind of value from getting it perfect in six weeks.
When is this the wrong way to spend the money?
Three cases, said plainly, because knowing them saves you more than the price difference.
A complex product. Zugo does not replace a development team when the thing you are building is a real application with unusual rules. It will get you to a working, published version quickly, and then the sensible move is GitHub export and a developer. The project is yours, so that door stays open by design.
A site whose value is the content. If the copy is unwritten and the photographs unshot, the build is not your constraint and no pricing tier fixes that.
Very specific business logic. Approval flows, odd permission matrices, and heavy edge cases arrive through edits rather than one prompt. That works, and it is priced per edit, so an unusual product costs more than a conventional one. Knowing that upfront is better than discovering it at edit twenty.
For the common case, a single person who needs a credible site or a small app published soon, the arithmetic above is the whole story: a few dollars of actions, an evening of your time, and a live address at the end of it. You can run the numbers on your own project with the free credits at zugo.dev, and if the timeline matters as much as the price, see how long a build actually takes.