ScraperAPI vs Bright Data (2026): 100 sites compared

How we tested

The October 30, 2026 run put 17 providers against the same 100 sites, five attempts each, 500 requests per provider, with a 90-second timeout. An attempt counts as a pass only when the response contains a marker string unique to the real page, so a served CAPTCHA, a consent wall, or a 200 with an empty body all count as failures. The full method and the raw numbers are on the benchmark page.

We build String and we rank first in this run. That is a conflict, and the answer to it is that the harness is public and the target list is published, so you can run it against your own domains. The number that matters to you is your own list, not ours.

Two notes on configuration belong here rather than in a footnote.

The ScraperAPI adapter sends ultra_premium=true with caching off and rendering off, and takes raw HTML back. That is ScraperAPI's advanced-bypass tier, the strongest request it publishes short of adding a browser. ScraperAPI's credit documentation prices it at 30 credits a request, and 75 with render=true. So the ScraperAPI number on this page is what its top tier does, at its top-tier price.

The Bright Data adapter calls the Web Unlocker /request endpoint with raw output and a US country, and no custom headers or cookies. Web Unlocker picks the network, fingerprint and retries itself. Its feature documentation lists a few opt-in controls, such as forcing a browser render and premium domains, and the run used none of them. Bright Data's documentation claims a 98% success rate. Its pricing page FAQ says the unlocker succeeds at "an extremely high success rate (typically 100%)". We measured 75.4% on this target list.

Where the gap appears

The aggregate hides the shape. In the October 30, 2026 run, full passes were close: ScraperAPI cleared 52 sites at 5 of 5, Bright Data 49. Shutouts were close too.

Both vendors fail the same way: about two sites in five return some pages and not others.

That matters more than it looks. A site that returns 3 of 5 does not announce itself. It fills your table with rows, and the missing 40% shows up later as a gap in a price history or a hole in a competitor feed. Retries close the gap and raise the bill, and because both vendors bill per successful request, the retries that succeed are the ones you pay for.

There is one difference in how they fail. Of ScraperAPI's 118 failed attempts, 23 came back with a 2xx status but without the target page's content. Of Bright Data's 123 failures, 6 did. A 2xx response that is not the page is the hardest failure to catch in code, and on ScraperAPI it can also bill, because its status-code documentation charges any 200 response.

Anti-bot coverage

The two providers are not weak in exactly the same places, which is the swap set from a different angle. Success rates are from the October 30, 2026 run.

Akamai is ScraperAPI's clearest win, 9 points across 20 sites. Kasada is where both are weakest, and Bright Data less so. DataDome and the proprietary group are effectively tied. Read the row that matches whatever protects your targets, because the aggregate will mislead you in both directions.

Speed, where they overlap

Latency figures here are from the October 30, 2026 run. Comparing mean latency across all attempts rewards a provider for failing quickly, so the useful comparison is restricted to the 29 sites both providers cleared 5 of 5. On those, neither side has a timeout skewing the number.

ScraperAPI's mean was 6.46 seconds and its median 3.90 seconds. Bright Data's mean was 13.14 seconds and its median 10.00 seconds. Bright Data was at least twice as slow on 15 of the 29 sites, and at least three times as slow on 11. ScraperAPI was the slower of the two on 8.

Across the whole run, Bright Data's per-target latency median was 23.0 seconds and its 95th percentile 75.7 seconds, against a 90-second timeout. ScraperAPI's were 6.5 and 46.0. On the benchmark's failure-aware latency score, ScraperAPI ranked sixth of 17 at 15.19 seconds and Bright Data fifteenth at 27.88.

This is a genuine trade and not a defect. Bright Data is slower partly because it retries harder, and the retrying is what buys its social and real estate sites. If your job runs nightly and finishes by morning, 23 seconds a page costs you nothing. If a user is waiting on the response, it rules Bright Data out for that path.

What ScraperAPI is

Credit figures in this section are from ScraperAPI's credit documentation, checked November 9, 2026. ScraperAPI is a request-level scraping API owned by saas.group, which announced in April 2026 that it had acquired Traject Data. You send a URL and get HTML back. Its shape is a credit menu: a normal page is 1 credit, Amazon, Walmart and eBay are 5, Google and Bing are 25, LinkedIn is 30, and a Cloudflare, DataDome, PerimeterX or Akamai bypass adds 10. Modes are opt-in flags you set per request: premium and render cost 10 credits each, both together 25, ultra_premium 30, and ultra_premium with rendering 75. A per-request max_cost ceiling, described in its cost-control docs, stops a request before it spends more than you set. It also ships structured-data endpoints, an AI parser, a crawler, the no-code DataPipeline, and an MCP server.

What Bright Data is

Per Bright Data's site, checked November 9, 2026, its Web Unlocker API is one endpoint on top of a very large proxy business: 400M+ residential IPs, ISP and datacenter pools, Scraper APIs for 1,200+ websites, datasets from 900+ domains, and a managed acquisition service. By default the Unlocker picks the network, sets the fingerprint, solves the challenge, retries, and charges you only when a page comes back. Its overview says it does not drive a browser, pointing scripted actions to the Browser API, and that search engines belong to a separate SERP API.

One boundary is worth stating precisely, because it is easy to misread. Bright Data's docs say social network account management is not a supported use case, naming Facebook, Instagram, X, LinkedIn and others. That is about operating accounts, not about reading public pages. The Unlocker returned public pages on Instagram, Facebook and X in this run, on all five attempts each.

Pricing, and what the tested path costs

Neither vendor has a plan near $100 a month, so there is no clean like-for-like anchor. Here is what each publishes, as checked on its pricing page on November 9, 2026.

ScraperAPI's Hobby plan is $49 a month for 100,000 credits, which is $0.49 per 1,000 credits. Startup is $149 for 1,000,000 and Business is $299 for 3,000,000. Because a credit is not a page, the page price depends on the target and the mode:

Credits do not roll over: ScraperAPI's own FAQ says the balance "resets when your subscription renews". Two limits catch people out below $299: geotargeting on Hobby and Startup is US and EU only, and pay-as-you-go past your allowance is not available until the $475 Scaling plan.

Bright Data's Web Unlocker pricing has a free tier of 5,000 requests a month, pay as you go at $1.50 per 1,000 requests with no commitment, and a Scale plan at $499 a month that includes 383,000 requests, then $1.30 per 1,000. The page also offers to match a first deposit up to $500. One rate is not public: premium domains, the sites Bright Data classes as harder to unblock, bill at a higher rate that the docs say appears in the zone's estimated cost. If your list includes them, the public rate is not your rate.

Set those side by side on the configuration we tested. ScraperAPI's ultra_premium request costs $14.70 per 1,000 on Hobby and $4.47 on Startup, against Bright Data's $1.50 pay as you go. On that path Bright Data is the cheaper vendor until ScraperAPI's Business plan, and even there it is $2.99 against $1.50. The comparison flips on cooperative pages: ScraperAPI's 1-credit request is $0.15 per 1,000 on Startup, a tenth of Bright Data's rate. Which number applies depends on how many of your sites need the bypass.

Both vendors bill only successful requests, with caveats. ScraperAPI's docs charge for 200 and 404 responses, plus requests you cancel before 70 seconds; a CAPTCHA page returned as 200 can therefore cost credits until ScraperAPI adds it to its detection list. Bright Data's docs say you are "charged only for successful requests to your target domain", except that turning on custom headers, cookies or expect elements switches the zone to billing 100% of requests. So neither normal rate needs dividing by its success rate. The benchmark failures mostly cost you time, not money.

For a third reference point, String's Growth plan is $99 a month (as of October 2026), all of it usage, at $0.16 per 1,000 standard fetches, $0.90 per 1,000 premium-proxy fetches and $3.25 per 1,000 premium browser requests. Unused credit rolls over, blocked requests cost nothing, and the first 20,000 standard fetches are free with no card.

Which one to pick

Industry figures below are from the October 30, 2026 run, and prices are as checked on November 9, 2026.

Pick Bright Data if your list includes social platforms, real estate, review sites or pharmacy pages. Its industry cuts were 87.5% on social media, 94.3% on real estate, 93.3% on reviews and local, and 84.0% on grocery and food, against ScraperAPI's 60.0%, 85.7%, 70.0% and 72.0%. It is also the cheaper of the two on hard pages at its $1.50 pay-as-you-go rate. Accept the latency, and expect partial results on about two in five sites.

Pick ScraperAPI if your targets skew toward travel, news and finance, ticketing or jobs, or if speed matters. It was 86.7% on travel, 88.9% on news and finance, 80.0% on tickets and events, and 88.0% on jobs and hiring, against Bright Data's 68.9%, 71.1%, 45.0% and 76.0%. On the pages both can reach it is roughly two and a half times faster at the median, and on cooperative pages that stay at 1 credit it costs far less.

Pick neither if your list spans both groups. This is the case the swap set exposes. Five sites returned only through Bright Data and three only through ScraperAPI, on the strongest configuration we could send each. Running both vendors works and some teams do it, but it means two contracts, two integrations and two failure modes.

For the record on the same run, String returned 5 of 5 on all 8 of those sites, and 98.6% (as of October 2026) overall with a 4.24-second latency score. It was not perfect. It returned 4 of 5 on yellowpages.com and marketwatch.com, where both ScraperAPI and Bright Data returned 5 of 5. On barrons.com, String and Bright Data returned 4 of 5 to ScraperAPI's 5. On kroger.com, String returned 2 of 5, ScraperAPI 5 and Bright Data 3. On skyscanner.net all three returned some attempts and not others. Those are the losses, named. You can check any of it against the harness or start on the free tier.

FAQ

ScraperAPI vs Bright Data, which should I pick?

It depends on your target list, not on which is better overall. In the October 30, 2026 run of the Web Data Frontier Benchmark the two were nearly tied, 76.4% against 75.4% across 100 sites. Bright Data reaches social, real estate and review sites that ScraperAPI does not, and ScraperAPI reaches ticketing, finance and travel sites that Bright Data does not. On 8 of the 100 benchmarked sites only one of them worked, so check which of the two exclusive lists overlaps your domains before deciding on the headline number.

Which one is more reliable?

Neither, by enough to matter. In the October 30, 2026 run of the benchmark, ScraperAPI returned nothing at all on 9 sites and a partial result on 39. Bright Data returned nothing on 11 and a partial result on 40. The difference is in how a failure looks: 23 of ScraperAPI's 118 failed attempts came back with a 2xx status but not the real page, against 6 of Bright Data's 123, so ScraperAPI's failures are harder to detect in code.

Which one is faster?

ScraperAPI, on the pages both can fetch, in the October 30, 2026 run of the benchmark. Across the 29 sites both cleared on all five attempts, ScraperAPI's median was 3.90 seconds and Bright Data's was 10.00. Bright Data was at least twice as slow on 15 of those 29. On the failure-aware latency score, ScraperAPI ranked sixth of 17 providers and Bright Data fifteenth.

Do either of them charge for blocked requests?

Normally no, per both vendors' docs checked November 9, 2026. ScraperAPI charges only for 200 and 404 responses, plus requests cancelled before 70 seconds, though a CAPTCHA page returned with status 200 can still use credits. Bright Data charges only for successful requests, unless you enable custom headers, cookies or expect elements, which bills every request.

Is Bright Data cheaper than ScraperAPI?

On protected pages, usually yes, on prices checked November 9, 2026. The benchmark sent ScraperAPI's ultra_premium request at 30 credits, which works out to $14.70 per 1,000 pages on the $49 Hobby plan and $4.47 on the $149 Startup plan. Bright Data's Web Unlocker is $1.50 per 1,000 pay as you go, or $1.30 past the 383,000 requests in its $499 Scale plan, before any premium-domain surcharge. On cooperative pages that stay at 1 credit, ScraperAPI Startup is $0.15 per 1,000 and far cheaper.

Which configuration of ScraperAPI did the benchmark test?

Its ultra_premium advanced-bypass tier, with rendering and caching off. ScraperAPI prices that at 30 credits a request (credit documentation, checked November 9, 2026). The default 1-credit request was not tested in this run, so this page has no success figure for it. Test it on your own URLs with max_cost set before assuming the cheaper path returns the same pages.

Which handles Cloudflare and DataDome better?

They are close. In the October 30, 2026 run of the benchmark, Bright Data returned 74.7% across the 15 Cloudflare-protected sites against ScraperAPI's 70.7%. On the 19 DataDome sites, ScraperAPI returned 75.8% and Bright Data 74.7%. The clearer gaps are elsewhere: ScraperAPI leads on Akamai, 81.0% against 72.0% across 20 sites, and Bright Data leads on Kasada, 66.7% against 53.3% across 3.

Can I verify these numbers myself?

Yes. The target list, the pass rule and the per-site results are published on the benchmark page, the per-attempt data is in the public run file, and the run is described in the October 2026 write-up. We build String and we rank first in the run, so the point of publishing the harness is that you do not have to take the ranking on faith. Run it against your own domains.

Sources

Checked November 9, 2026.

• ScraperAPI pricing
• ScraperAPI credits and request costs
• ScraperAPI supported parameters
• ScraperAPI cost control
• ScraperAPI API status codes
• ScraperAPI premium residential and mobile pools
• ScraperAPI geotargeting
• ScraperAPI LLM integrations
• ScraperAPI MCP server
• ScraperAPI structured data endpoints
• ScraperAPI DataPipeline
• ScraperAPI Crawler
• ScraperAPI acquires Traject Data
• Bright Data Web Unlocker pricing
• Bright Data Web Unlocker API overview
• Bright Data Web Unlocker quickstart
• Bright Data Web Unlocker configuration
• Bright Data Web Unlocker features
• Bright Data Web Unlocker error codes
• Bright Data Web Unlocker product page
• Bright Data SERP API
• Bright Data Browser API
• Bright Data residential proxies
• Bright Data Scraper APIs
• Bright Data datasets
• Benchmark ScraperAPI adapter
• Benchmark Bright Data adapter
• October 30, 2026 run file
• String pricing
• String Web Access API
• Web Data Frontier Benchmark
• October 2026 benchmark write-up
• Why the web scraping benchmark problem is hard
• String vs ScraperAPI
• String vs Bright Data
• Firecrawl vs Bright Data for AI agents
• ScraperAPI alternatives
• Bright Data alternatives
• Web scraping API pricing compared