Pay for the pages you process.

The unit every plan is sold in. One successful, uncached API call against one URL spends credits according to what that call costs us to run: a page fetch (/html, /markdown, /meta, /screenshot, /console) is 1 credit, a Lighthouse audit is 2, /keywords is 15, /summarize is 17 and /extract is 24. A /scrape request spends the sum of the operations it runs. The one extra is residential: true, which adds 25 credits per page visit — once per visit, so a /scrape pays it once however many operations it runs.

Because they don't cost us the same, and pricing them as if they did meant charging light users for work they never do. A page fetch runs on hardware we already own, and so do a markdown conversion and a metadata read — they read the page we already rendered. The AI endpoints are a bill from a model provider that scales with how much of the page the model reads and writes back. The credit weights are those costs, rounded.

No. When you ask for the same URL and operation again within your freshness window, we return the cached result — instantly, and without spending a credit. A repeat of a request that is still running waits for it, free as well. You set the window per request with max_age (default 7 days). It is the cheapest lever you have on your bill.

They spend nothing. That covers a target that fails — DNS failure, connection refused, timeout, bad certificate, a 404 or a 5xx — a bot check we couldn't get past, a page the site's robots.txt disallows, and a residential request no exit was free for. You're only billed for work that produced a result.

Every response carries an X-Quota-Cost header saying what that request cost, alongside X-Quota-Remaining, X-Quota-Overage and X-Quota-Reset. In the dashboard, the Billing screen itemises the month by operation — calls, credits each, credits total and share of the bill — so you can always see which endpoint is spending your allowance.

On a paid plan, nothing stops — ever. Your calls keep being served, uncapped, and every credit past the allowance costs $1.50 per 1,000 credits, totalled when the month closes and billed as an invoice of its own. There is no ceiling to hit and no request that gets refused for want of credits. On the Free plan, which has no card on file to bill the extra to, the endpoint returns HTTP 429 with a JSON body naming your limit, what you've used and how many credits the refused request needed.

How many requests you can run at the same time: 1 on Free, 3 on Starter, 8 on Pro, 20 on Scale. Credits are how much work you can do in a month; this is how fast you get through it. A crawl of 5,000 pages costs the same credits however quickly you run it — running 20 in parallel instead of one at a time is the difference between minutes and hours. Go over and the request comes back with a 429 and error concurrency_limit; nothing is spent, and you retry as soon as one of your own finishes. Cache hits don't take a slot at all.

Yes, on every paid plan, with as many members as you need. Invite people by email from Organization settings: everyone works on the same projects and draws on the same monthly credits, and admins look after the plan, billing and the team. The Free plan is for one person. One login can also belong to several organizations — one per client or company, each with its own plan and bill — and switch between them from the account menu.

Three ways, none of which need you to go looking. Every API response carries an X-Quota-Overage header counting the credits you've spent past your allowance. We email your organization admins the first time in a month you cross. And the app shows a banner with the running total, for as long as you're over. You should never learn about overage from an invoice.

For a spike, yes; as a habit, no, and deliberately so. Overage is priced above the credits your plan already includes, so a month that goes a little over costs a few dollars, while traffic that has permanently outgrown your plan costs more in overage than the next plan up costs outright. The Billing screen points that out while it is happening.

Any time, in both directions, from Billing in your dashboard. A change takes effect straight away and Stripe prorates it: moving up charges the difference for the rest of the period, and moving down credits the unused part of the old plan against your next invoice.

You pay ten months up front and get twelve — two months free. The monthly credit allowance is unchanged; annual is about the price, not the allowance.

No. Your allowance resets on the 1st of each month, UTC. Unused credits don't accumulate — which is also why we don't charge for cache hits, so steady usage costs less than the raw call count suggests.

From Billing in your dashboard, which opens Stripe's billing portal. Cancelling stops the renewal; your plan stays active until the period you've paid for ends, then drops to Free. Everyone on your team keeps their access; inviting anyone new needs a paid plan again.

No — every price here is in US dollars and before tax. Customers in the EU pay their own country's VAT on top, added at checkout. A business with a valid VAT number in an EU country other than Spain is reverse-charged and pays none; enter it at checkout. Outside the EU nothing is added. Extra credits past the allowance are taxed the same way.

Cards, via Stripe. We never see or store your card details. If you need to pay by invoice or bank transfer, email us and we'll arrange it.