ScraperAPI Pricing and Cost per 1,000 Pages

ScraperAPI is one of the cheapest ways to fetch a page on the internet, and one of the easiest bills to underestimate. The confusion is not hidden pricing. Every multiplier is published, which is more than several vendors in this category manage. The confusion comes from the fact that the plan page sells credits and your pipeline consumes pages, and the exchange rate between the two moves by a factor of 75 depending on what you turn on.

Every figure below was read from the ScraperAPI pricing page and its credits and requests documentation on 6 September 2026. No aggregator numbers, because the aggregators covering this category are quoting a five-tier ladder that ScraperAPI no longer publishes.

ScraperAPI pricing tiers in 2026

Six self-serve tiers and a quoted Enterprise plan. What changes between them is the credit allowance, the concurrency ceiling and, crucially, whether you are allowed to spend past your allowance at all.

Read the right-hand column downward and the volume curve is steep at the start and almost flat after it. Hobby to Startup cuts the unit price by 70 percent. Startup to Business cuts it by another third. After Business, the four remaining tiers sit between $0.100 and $0.092 per 1,000 credits, a spread of eight percent across a six-fold jump in price. Above three million credits you are no longer buying cheaper pages. You are buying concurrency and, as the next section explains, permission to overspend.

What happens when you run out of ScraperAPI credits

This is the mechanic that catches people, and it is documented plainly in the ScraperAPI pricing FAQ rather than buried. Pay-as-you-go overage, which lets you keep scraping past your allowance at a fixed per-credit rate with a spending cap you set, is available on the Scaling, Professional, Advanced and Enterprise plans only.

On Hobby, Startup and Business, hitting 100 percent of your credits before renewal does not produce a bigger invoice. It produces a stop. Your two documented options are to upgrade to the next plan mid-cycle or to contact support about a custom plan. There is no third option where the pipeline quietly keeps running and you settle up later.

Pair that with the second policy in the same FAQ: unused credits do not roll over, and your balance resets when the subscription renews. Put the two together and the three cheapest plans push you toward buying more than you need. You cannot bank a quiet month against a busy one, and you cannot spill over in the busy one, so the only safe sizing is your peak month, every month. That is not a criticism of the pricing so much as a fact you should model, because it is the difference between the $149 plan looking cheap and the $299 plan being the one you actually have to buy.

How many credits does a ScraperAPI request cost?

One credit buys a standard page fetch. Everything else is a published multiplier, and the target domain has its own price independent of the parameters you send.

A useful detail sits on the other side of that table. ScraperAPI documents a list of parameters that cost nothing extra: wait_for_selector, country_code, session_number, device_type, output_format, keep_headers and autoparse. Geotargeting and sticky sessions being free is genuinely unusual, and autoparse being free means the structured output it does support for certain sites is not an upcharge. If a country code or a session is all that stands between you and the page, you are still paying one credit.

ScraperAPI also bills only for successful requests, defined as 200 and 404 status codes, plus requests you cancel before giving the service 70 seconds to finish. A 403 or a timeout you waited out costs nothing. On defended targets, where a double-digit failure rate is normal, that policy is worth real money and it is one of the clearest reasons to keep ScraperAPI for the hard half of a workload.

ScraperAPI cost per 1,000 pages, by plan and by page type

Multiply the credit rate by the credits per page and you get the only number that matters for a budget. This table is the pricing page translated into pages.

The left column is why ScraperAPI has the reputation it has. Ten cents per thousand server-rendered pages is close to unbeatable, and if your targets are ordinary HTML documents you should probably stop reading and go buy the Business plan.

The second column is where most 2026 workloads actually land. Modern documentation sites, marketing pages, job boards, product catalogs and help centers build their content in the browser, so render=true is not an optional extra, it is the default. That moves Hobby from $0.49 to $4.90 per 1,000 pages, a ten-fold jump that people discover in week three rather than week one. The same jump is why the Hobby plan, read as pages rather than credits, is 100,000 plain pages or only 10,000 rendered ones.

How many pages does each ScraperAPI plan actually buy?

Same arithmetic, expressed the way a capacity plan needs it.

Note the bottom right cell. The $1,975 plan buys 286,666 pages a month if every page needs a browser and ultra premium routing. The same $1,975 buys 21.5 million pages if none of them do. Choosing the plan is easy. Knowing which column you live in is the whole job, and the only way to know is to run a sample of your real URLs and read what each one billed.

Does ScraperAPI have a free plan?

Yes, two things that are easy to confuse. There is a 7-day trial that grants 5,000 API credits with no credit card required, and there is a standing free plan of 1,000 API credits with a maximum of 5 concurrent connections. The trial is for evaluating throughput. The free plan is for keeping a small script alive. Neither is enough to benchmark a production workload, which is fine, because 5,000 credits is more than enough to answer the only question that matters before you pay: what does one of my pages actually cost.

ScraperAPI also publishes a 7-day no-questions-asked refund policy and a Domain Cost Estimator in the dashboard that reports the credit cost for a specific URL before you scrape it, plus a max_cost parameter that caps what any single request may spend. Those three together make the measurement exercise below cheap and reversible.

How to work out what ScraperAPI will really cost you

Do not model this from the pricing page. Model it from your own URLs, in about twenty minutes.

1. Take 200 URLs that represent your actual crawl, not 200 homepages. Include the awkward ones.
2. Run them on the free credits with no flags set. Count how many return the content you need in the raw HTML. That share is your 1-credit population.
3. Re-run only the failures with render=true. Anything that now works is a 10-credit page. Anything still empty needs premium or ultra premium, at 25 or 75 with rendering.
4. Check the domain list separately. Any Amazon, Walmart or eBay URL is 5 before parameters, any Google or Bing URL is 25, any LinkedIn URL is 30.
5. Compute a weighted average credits per page, multiply by your monthly page volume, and pick the tier whose allowance covers your peak month, because credits do not roll over and the three cheapest plans will not let you spill past the line.

Teams that skip step 5 are the ones who find their crawler halted on the 22nd. If you want the general version of that arithmetic across vendors rather than just this one, we wrote up how web scraping API credits work separately.

Where ScraperAPI is the right buy, and where it is not

ScraperAPI is proxy infrastructure with a very good price at volume. If your targets are ordinary permitted pages that return their content as HTML, nothing in this category beats ten cents per thousand, and the free-on-failure policy plus free geotargeting make the hard targets more affordable than the sticker suggests. That is a real product and this is not an argument against buying it.

What you get back, though, is a page. Raw HTML, or a screenshot, or autoparsed output for the specific sites ScraperAPI has built parsers for. If the destination is a vector store, a knowledge base or a typed record in your database, the boilerplate stripping, the markdown conversion and the schema layer are still your code to write and maintain. That work does not appear on any pricing page and it is usually the part that keeps costing after launch.

The LinkedIn row is worth one more sentence for the same reason. At 30 credits a page it is the most expensive line in the table, which is why recruiting pipelines built on scraped profiles get expensive faster than anyone plans for, and why most teams eventually narrow the funnel earlier and let an automated first-round screen do the filtering instead of buying a bigger crawl.

ClawEngine is priced the other way around, and it is worth saying where that loses. We bill a flat monthly page allowance where one crawled, rendered, schema-typed page counts as one page: $39, $99 and $399 a month for roughly 50,000, 250,000 and 1,500,000 pages, which is $0.78, $0.40 and $0.27 per 1,000. That pricing is published but not yet open for purchase. Against ScraperAPI Business at $0.10 per 1,000 plain pages, we are five to eight times more expensive on server-rendered HTML and we always will be, because we render everything. Against the same plan at $1.00 per 1,000 rendered pages, we are cheaper from the $99 tier upward, and the extraction is included rather than being code you write. We also defeat no anti-bot system, run no residential pool and do not touch logins, so every ultra premium row in that table is a job we simply cannot take.

If you are shopping this decision more broadly, our ScraperAPI alternatives comparison puts eleven of these tools side by side on cost and output format, and the Olostep alternatives page covers the opposite failure mode, a vendor with a very low headline rate that does not publish a per-request credit table at all.

Frequently asked questions about ScraperAPI pricing

Do ScraperAPI credits roll over?

No. ScraperAPI states that your credit balance resets when the subscription renews, and that you can move up or down a plan at any time from the billing page if usage changes. Nothing carries forward, so a quiet month is money spent rather than credit banked. That is the standard model in this category, but it interacts badly with the missing pay-as-you-go on the three cheapest plans, and together they are the argument for sizing on your peak month.

What happens if I exceed the concurrent thread limit?

You get a 429 asking you to bring concurrency back within your plan limit and retry. There is no extra charge and no queue: the request is turned away rather than held. That is worth knowing because it changes how you handle backpressure. A 429 from ScraperAPI is a client-side scheduling problem, not a billing event, so retry with a token bucket sized to your thread count rather than treating it as a failure.

Is the Enterprise discount applied to every request type?

Almost. ScraperAPI says custom Enterprise pricing applies across your usage with one documented exception: requests to Ultra Premium carry their own minimum rate. If your workload leans on the hardest targets, that carve-out is the first thing to raise with sales, because it applies to exactly the 30 and 75 credit rows where a discount would be worth the most.

Which ScraperAPI plan should I start on?

Start on the free 5,000-credit trial and measure, then jump to whichever tier covers your peak month, and skip Hobby unless your volume is genuinely tiny. Hobby is $0.49 per 1,000 credits against Startup at $0.149, so the second tier is three and a bit times cheaper per unit for three times the money. If you expect to render most pages, Hobby buys 10,000 pages a month, which almost no production pipeline is.

How does ScraperAPI compare with a flat per-page crawl API?

The break-even is entirely about how many of your pages need a browser. On ScraperAPI Business, a plain page is $0.10 per 1,000 and a rendered one is $1.00. A flat per-page plan charges the same for both, so it loses badly on static HTML and wins once rendering is the norm rather than the exception. Count the share of your target pages that return empty without JavaScript. If it is under about a third, credits are cheaper. If it is most of them, which for modern sites it usually is, flat pricing is both cheaper and easier to forecast.