NotCheap MAKE EVERY CREDIT COUNT
Analyze Your Costs

Apify vs Bright Data vs ScrapingBee: Real Web Scraping Cost per 10 Million Pages in 2026

Share this guide

The short answer: the three vendors bill different things, so the same 10 million pages can cost about $1,600 or about $61,000 on Apify depending on page type, retries and proxy bandwidth, while Bright Data's Web Unlocker charges a flat $1.30–$1.50 per 1,000 successful requests ($13,000–$15,000) and ScrapingBee prices in credits whose multipliers (5x for JavaScript rendering, 25x with a premium proxy, 75x for stealth) turn 10 million pages into anywhere from 10 million to 750 million credits.

For static pages with a high success rate, Apify's compute-based billing is by far the cheapest modeled route (about $1,600 on the Scale plan, before a datacenter proxy cost that is not publicly disclosed). For JavaScript-heavy, anti-bot pages that need residential proxies and retries, the order inverts: Apify's bill is dominated by residential bandwidth ($7.50 per GB across roughly 7,100 GB of attempts), Bright Data's success-based Unlocker is the lowest published all-in price, and ScrapingBee's credit multipliers push it past its largest published plan into custom enterprise territory. "Pages fetched" is not a comparable unit across these platforms until the page type, retries and bandwidth are fixed.

The catch is that the cheapest option flips depending on what you're scraping. For static pages with a high success rate, Apify's compute-based billing is by far the cheapest modeled route (about $1,600 on the Scale plan, before a datacenter proxy cost that is not publicly disclosed). For JavaScript-heavy, anti-bot pages that need residential proxies and retries, the order inverts. Apify's bill is dominated by residential bandwidth ($7.50 per GB across roughly 7,100 GB of attempts), Bright Data's success-based Unlocker is the lowest published all-in price, and ScrapingBee's credit multipliers push it past its largest published plan into custom enterprise territory.

My practical rule of thumb: "pages fetched" is not a comparable unit across these platforms until you've pinned down the page type, retries and bandwidth. Get those three numbers first, then compare prices. Comparing headline rates before that is how people end up with a surprise invoice.

The billing-unit mismatch: why headline rates deceive

Before running benchmarks, it helps to translate each platform's pricing model into real engineering units:

  • Apify bills for serverless compute and bandwidth: You pay for Actor compute units (1 Compute Unit = 1 GB RAM allocated for 1 hour) plus proxy traffic per gigabyte. If your crawler hits anti-bot blocks, retries failed requests, or renders heavy JavaScript bundles, every wasted attempt consumes both compute and expensive residential bandwidth.
  • Bright Data bills for guaranteed outcomes: Its Web Unlocker and Scraping APIs charge exclusively for HTTP 200 responses. Proxy rotation, browser emulation, CAPTCHA bypasses, and failed retries are absorbed by Bright Data without inflating your invoice.
  • ScrapingBee bills for feature-weighted API credits: A plain curl-style request without JavaScript costs 1 credit. Turning on headless browser rendering costs 5 credits. Adding premium residential proxies costs 25 credits. Turning on full anti-scraping stealth mode costs 75 credits. A 10-million-page crawl can therefore require between 10 million and 750 million credits.

Platform pricing breakdown

Apify

Apify charges tiered subscription commitments that act as prepaid usage credits:

  • Free: $0 with $5 included usage
  • Starter: $19/month (prepaid usage credit)
  • Scale: $199/month
  • Business: $999/month
  • Enterprise: Custom SLA and dedicated infrastructure

Platform unit rates: Compute units cost $0.20 on Starter, $0.16 on Scale, and $0.13 on Business. Residential proxy bandwidth scales from $8.00/GB on Starter down to $7.50/GB on Scale and $7.00/GB on Business. Light HTTP crawlers (using Cheerio or Crawlee) typically consume around 0.95 compute units per 1,000 pages (~$0.19), while full headless Chromium browsers consume 3.35 to 4.5 compute units per 1,000 pages (~$0.67 to $0.90) before accounting for proxy traffic.

Bright Data

Bright Data provides both raw proxy network pools and managed extraction APIs:

  • Web Unlocker: $1.50 per 1,000 successful requests pay-as-you-go, or $1.30 per 1,000 on a $499/month commitment (380,000 requests included). This managed layer handles automatic IP rotation, header spoofing, CAPTCHA solving, and browser rendering with zero charge for failed attempts.
  • Residential Proxies: Around $3.54 to $4.00 per GB under monthly commitments ($499 for 141 GB), with pay-as-you-go rates higher and city-level targeting adding 20% to 40% surcharges.

ScrapingBee

ScrapingBee structures its plans around monthly credit buckets:

  • Freelance: $49.99/month for 250,000 credits (10 concurrent requests)
  • Startup: $99.99/month for 1,000,000 credits (50 concurrent)
  • Business: $249.99/month for 3,000,000 credits (100 concurrent)
  • Business+: $599.99/month for 8,000,000 credits (effective rate of $0.075 per 1,000 credits)
  • Enterprise: Custom volume plans above 8 million credits

Normalized cost comparison: 10 million pages

To get a fair, apples-to-apples comparison, we look at two distinct extraction workloads:

  • Scenario A (Static HTML scraping): Unprotected news, public blogs, or directory listings. Light HTTP fetching (no browser rendering), 95% success rate on first attempt, 100 KB payload per page.
  • Scenario B (Hardened anti-bot targets): E-commerce pricing or flight portals behind Cloudflare or DataDome. Requires headless browser rendering, residential proxy pools, and experiences a 70% raw success rate (requiring 14.29 million total attempts to secure 10 million successful pages), transferring 500 KB per attempt after stripping images and media assets.

Cost at the normalized workload

ScenarioApify (Scale)Bright Data Web UnlockerScrapingBee
A: static, datacenter, 95% successAbout $1,600 compute (10,000 CU); datacenter proxy cost not published$13,000–$15,000 (an all-in unlocker; more capability than static pages need)10 million credits at 1 credit per page: exceeds Business+ (8 million); QUOTE-ONLY. ILLUSTRATIVE effective average: $750
B: JS-heavy, anti-bot, residential, 70% successAbout $61,229 ($7,657 compute for 47,857 CU + $53,571 residential bandwidth for 7,143 GB)$13,000–$15,000 (success-based; retries and proxy bandwidth absorbed)250 million credits at 25 per page (premium proxy plus rendering): QUOTE-ONLY. ILLUSTRATIVE effective average: $18,750; stealth at 75 credits: $56,249
Web scraping cards at 10M pages: Bright Data Web Unlocker at $13k-$15k, ScrapingBee 8M plan at $599.99, and Apify compute from $1,600 to $61,229.
Cost model comparison across web scraping architectures at 10M monthly pages drawn from the scenarios analyzed above.

Why proxy bandwidth—not compute—explodes self-managed costs

The big cost gap in Scenario B points to the most common way scraping budgets go sideways: residential bandwidth leaks.

On Apify, container compute for 14.29 million attempts is relatively modest at $7,657. But residential proxy traffic is billed at $7.50 per GB. At 500 KB per request across 14.29 million attempts, your crawler downloads 7,143 GB of network data, which comes to $53,571 in proxy bandwidth fees alone. That's 87.5% of your entire invoice.

When you run self-managed scrapers, every failed request, 403 Forbidden page, or CAPTCHA challenge burns full residential bandwidth. Bright Data's Web Unlocker works differently: it charges $1.30 to $1.50 per 1,000 successful responses and absorbs all 4.29 million failed attempts, proxy rotations, and bandwidth under that flat fee. That cuts total extraction spend from $61,200 down to $13,000.

In real-world production, this is the part that catches teams off guard. Compute is easy to estimate and rarely the problem, while bandwidth on retries is what quietly eats the budget. If your targets are protected, price the retries before you price anything else.

Sensitivity analysis: your four biggest cost levers

  1. Payload size optimization: In Scenario B on Apify, dropping the downloaded payload from 500 KB to 250 KB (by aggressively aborting unnecessary fonts, CSS, and analytics scripts) cuts proxy bandwidth from $53,571 to $26,785, saving over $26,000 immediately.
  2. First-pass success rate: Self-managed crawlers scale costs proportionally to 1 / (success rate). Raising target success from 70% to 90% reduces required attempts from 14.29 million to 11.11 million, saving $12,000 in bandwidth and compute.
  3. Credit multipliers on API scrapers: Moving from ScrapingBee's premium proxy mode (25 credits) to stealth mode (75 credits) triples your credit consumption from 250 million to 750 million credits.
  4. Subscription tiers on Apify: Moving from Starter to Business drops compute from $0.20 to $0.13/CU and residential bandwidth from $8.00 to $7.00/GB, saving roughly $5,000 a month on high-volume runs.

Practical architecture guide: which tool to use

  • Choose Apify when: Your target sites are static or moderately protected, your engineering team enjoys building custom Crawlee/Playwright pipelines, and you need scheduled background data pipelines writing directly to cloud datasets.
  • Choose Bright Data Web Unlocker when: You are scraping enterprise e-commerce portals, social networks, or search engines with aggressive bot defenses. Paying $1.30–$1.50 per 1,000 successful pages is far cheaper and less stressful than maintaining a residential proxy rotation setup in-house.
  • Choose ScrapingBee when: You want a clean, reliable REST API that returns rendered HTML or JSON with simple query parameters, and your monthly volume fits comfortably within 1 to 20 million pages without you managing browser clusters.

Methodology and sources

  • Apify rates reflect confirmed platform pricing as of late 2026 for Starter ($19), Scale ($199), and Business ($999).
  • Bright Data figures reflect verified pay-as-you-go and committed plan tiers for Web Unlocker and residential proxy traffic.
  • ScrapingBee credit structures and plan tiers reflect verified documentation for Freelance through Business+ plans.
PASS IT ON

Useful for your team?

Share this guide with a colleague.