ScrapingBee turns JavaScript rendering on unless you turn it off, and a rendered request costs five credits. So the credit count on every plan card is five times its default request count. The $19 plan advertises 75,000 credits, which is 15,000 default requests, or exactly 1,000 through stealth proxies. This page converts every published tier into money per thousand real requests, from figures read on 9 September 2026.
We sell competing proxies and scraper endpoints, so weigh our reading accordingly. The weights below are theirs, quoted from their own documentation, and the division is ours.
The short version
Their documentation states that a headless browser is the default behaviour and costs 5 credits per request. Premium proxies with rendering cost 25 credits, and stealth proxies cost 75. The $19 Hobby plan therefore holds 75,000 plain requests, 15,000 default ones, or 1,000 stealth ones. Their FAQ also says they charge only for successful requests, then defines success as a 200, 404 or 410 status code.
What ScrapingBee publishes
These are the advertised plans, read from the pricing page on 9 September 2026. The fourth column is our own division, and it prices a credit rather than a request.
ScrapingBee plans and the price of one credit
| Plan | Per month | Credits | Per 1,000 credits | Concurrency |
|---|---|---|---|---|
| Free trial | $0 | 1,000 | free | not stated |
| Hobby | $19 | 75,000 | $0.253 | 25 |
| Freelance | $49 | 250,000 | $0.196 | 50 |
| Startup | $99 | 1,000,000 | $0.099 | 100 |
| Business | $249 | 3,000,000 | $0.083 | 200 |
| Business Plus | $599 | 8,000,000 | $0.075 | 400 |
scrapingbee.com pricing page, read 9 September 2026, with our own division
The ladder is gentler than most, at a little over three to one from top to bottom. The steepest single step sits between Freelance and Startup, where the credit price roughly halves.
Note also that prices are stated exclusive of VAT. Buyers in a VAT country should add their own rate before comparing against a gross quote.
Rendering is on by default
This is the line that changes every calculation, and their documentation states it plainly. ScrapingBee fetches the URL through a headless browser that executes the page JavaScript.
The same passage adds that this is the default behaviour and costs 5 credits per request. Setting render_js to false brings the request back to one credit.
Most rivals in this market work the other way round. A plain request is the default, and rendering is a switch you turn on knowing it costs more.
Here the expensive mode is the resting state. Anyone sizing a plan from the credit count without reading that sentence will over-estimate their capacity fivefold.
The single cheapest change available here costs nothing. Setting render_js to false on targets that need no browser divides the bill by five. Check which of your targets actually need rendering before you size anything.
The weights, and what they do to the money
Four weights are published, and the range between them is 75 to one.
Credits consumed per request
| Configuration | Credits |
|---|---|
| Plain request, render_js set to false | 1 |
| Default request, rendering enabled | 5 |
| Premium proxies without rendering | 10 |
| Premium proxies with rendering | 25 |
| Stealth proxies | 75 |
| Auto-Mode where every configuration fails | 0 |
scrapingbee.com documentation, read 9 September 2026
Multiply the credit price by the weight and the real rate appears:
Cost per 1,000 requests, by configuration and plan
| Configuration | Credits | Hobby $19 | Startup $99 | Business Plus $599 |
|---|---|---|---|---|
| Plain request | 1 | $0.25 | $0.10 | $0.07 |
| Default, rendering on | 5 | $1.27 | $0.50 | $0.37 |
| Premium, no rendering | 10 | $2.53 | $0.99 | $0.75 |
| Premium with rendering | 25 | $6.33 | $2.48 | $1.87 |
| Stealth proxies | 75 | $19.00 | $7.43 | $5.62 |
our own arithmetic on advertised figures, 9 September 2026
What the plan allowance is actually worth
Restate the same arithmetic as request counts and the plan cards read very differently:
Advertised credits, converted to requests
| Plan | Credits | Plain | Default | Premium rendered | Stealth |
|---|---|---|---|---|---|
| Hobby | 75,000 | 75,000 | 15,000 | 3,000 | 1,000 |
| Freelance | 250,000 | 250,000 | 50,000 | 10,000 | 3,333 |
| Startup | 1,000,000 | 1,000,000 | 200,000 | 40,000 | 13,333 |
| Business | 3,000,000 | 3,000,000 | 600,000 | 120,000 | 40,000 |
| Business Plus | 8,000,000 | 8,000,000 | 1,600,000 | 320,000 | 106,666 |
our own arithmetic on advertised figures, 9 September 2026
The Hobby row is the one to sit with. A plan advertising 75,000 credits holds exactly 1,000 stealth requests.
The free trial is 1,000 credits with no card. Against a defended target needing stealth, that is thirteen requests.
A mixed job shows the sizing problem clearly. Take 50,000 pages a month, where 80 percent are plain and 20 percent need premium proxies with rendering. That is 40,000 requests at one credit and 10,000 at twenty-five credits.
The total comes to 290,000 credits, which sits just above the Freelance allowance. Freelance at $49 falls short, so the job lands on Startup at $99. Fifty thousand pages for $99 is $1.98 per thousand, against a headline credit price of $0.099.
Three vendors, one definition of success
Their FAQ answers the failure question in a single sentence. They charge only for successful requests, returning with a 200, 404 or 410 status code.
Both halves of that sentence matter, and the second half is the one that costs money. A 404 is a successful request under this definition, billed at whatever weight it carried.
On stealth proxies, a dead link therefore costs 75 credits. On the Hobby plan that is one seventy-fifth of the entire month.
What makes this worth a section is that it is not one vendor's quirk. We have now read the same rule, in the vendors' own words, at three separate companies this week.
How three vendors define a successful request
| Vendor | Codes billed as successful |
|---|---|
| ScrapingBee | 200, 404 and 410 |
| ZenRows | 404 and 410 count as successful results |
| ScraperAPI | 200 and 404, plus requests you cancel before 70 seconds |
three vendor pricing pages and documentation, all read 9 September 2026
Three independent companies converging on the same definition suggests a market convention rather than a trick. It is still a cost, and it is still absent from every comparison table we have read.
Our earlier study measured the same weights across this market in why 250,000 credits is not 250,000 requests. The weights have not moved since August.
The one rule other vendors should copy
Auto-Mode is the exception to everything above, and it is a good one. Their documentation states that if every configuration fails, the request costs 0 credits.
That is a genuine no-result, no-charge rule, and it is rare. It applies to the mode that retries several configurations rather than to a single request.
It also sits oddly beside the 404 rule on the same site. One rule makes a run that produced nothing free, while the other bills a 404 at full weight.
The comparison, at an identical weight
Two vendors in this batch charge exactly 25 credits for a rendered request through premium proxies. That makes the comparison unusually clean, because only the credit price differs.
Cost per 1,000 rendered requests through premium proxies
| Where you buy it | Plan | Per 1,000 requests |
|---|---|---|
| ScrapingBee | Hobby, $19 | $6.33 |
| ScrapingBee | Startup, $99 | $2.48 |
| ScrapingBee | Business Plus, $599 | $1.87 |
| ZenRows | Launch, $57 | $5.70 |
| ZenRows | Scale, $456 | $2.28 |
| HProxy scraper API | no plan | published on our scrapers page |
two vendor pricing pages and our own arithmetic, 9 September 2026
Our unit is one successful call from a balance that does not expire, with no weight to decode. The full ZenRows conversion sits in the ZenRows teardown.
The minimum, the commitment and the expiry
The smallest paid plan is $19 a month, and the page states you can cancel anytime. No annual billing option appeared on the page we read.
The free trial is 1,000 credits with no card required. That is enough to test a parser and not enough to test a defended target.
Running out is handled by moving plans rather than by buying credits. Their FAQ describes upgrading temporarily and downgrading later, or renewing the subscription early from the dashboard.
Above 41 million credits a month, the FAQ directs buyers to contact them for a custom plan. No rate is published at that level.
Who this pricing suits
ScrapingBee suits people scraping plain pages who remember to disable rendering. At $0.07 per thousand plain requests on the largest plan, that rate competes with anything in this market.
It suits mid-volume defended scraping reasonably. Premium with rendering at $2.48 per thousand on the $99 plan is a fair number for what it absorbs.
It suits small volume against hard targets poorly. On Hobby, a stealth request costs $19.00 per thousand and the whole plan holds a thousand of them.
It also suits anyone who sizes a plan from the credit count poorly. The default weight is five, so the plan card overstates your capacity unless you act.
The same conversion for rival vendors sits in our ZenRows, ScraperAPI and Firecrawl teardowns.
What we could not verify
Neither the pricing page nor the FAQ says whether unused credits roll over monthly. On a credit plan that term decides the effective rate, and we make no claim about it.
No annual billing discount appeared on the page we read. Several rivals publish one, so its absence may be real or may simply be unlisted.
We found no published credit cost for screenshots or for their Google Search API. Both are listed as features without a weight attached.
The concurrency figure for the free trial is not stated. We recorded the paid tiers only.
Prices and weights both move. Every figure here was read on 9 September 2026, so check the live pages before committing.
Sources
- ScrapingBee, pricing page, read 9 September 2026. Plan names, monthly prices, credit allowances, concurrency limits and the VAT note.
- ScrapingBee, documentation, read 9 September 2026. The statement that a headless browser is the default behaviour and costs 5 credits per request.
- ScrapingBee, documentation, read 9 September 2026. Premium proxies at 25 credits with rendering and 10 without, and stealth proxies at 75 credits.
- ScrapingBee, documentation, read 9 September 2026. The Auto-Mode rule that a request costs 0 credits if every configuration fails.
- ScrapingBee, FAQ, read 9 September 2026. The statement that only successful requests are charged, defined as returning a 200, 404 or 410 status code.
- ScrapingBee, FAQ, read 9 September 2026. The handling when credits run out, and the custom plan threshold above 41 million monthly credits.
- ZenRows and ScraperAPI pricing pages and documentation, read 9 September 2026, for the two comparison rows and the definitions table.
- Our own arithmetic on those published figures, 9 September 2026, shown in every table above.
