> 16 Web Scraping APIs: Detailed 2026 Benchmark <

Plans are tiered as Free, Hobby, Pro, Business, Advanced, and Custom/Enterprise, billed in credits with concurrency rising from 5 (Free/Hobby) to 200 (Advanced). The headline gotcha is the 5x multiplier for JS render and 10x for residential (a residential plus render call burns 25 credits), plus per-domain surcharges that get applied automatically: 10 credits for most protected sites, 25 for g2, 30 for linkedin.

Credits don't map one-to-one to requests. Each feature stacks a multiplier on top of the base call:

• ·Datacenter (default): 1 credit
• ·Datacenter + JS render: 5 credits
• ·Residential / Mobile: 10 credits
• ·Residential + JS render: 25 credits
• ·Protected domain auto-upgrade (e.g. google, indeed, reddit): 10 credits
• ·G2: 25 credits
• ·LinkedIn: 30 credits

A few things to watch out for before signing up:

• !Auto-upgrade hides the real per-domain cost: requesting basic on capterra, google, indeed, reddit, tripadvisor, or trustpilot still bills 10 credits per call.
• !Residential proxy at 10 credits per call (25 with render) burns volume fast on hard targets.
• !Hard 2MB response cap on returned HTML.

The Web Scraping API starts at $19/month for 38K standard requests at $0.50/1K, with premium proxies plus JS rendering billed at $1.50/1K. A free tier offers 2K requests, and the popular $99 Pro plan unlocks 50 req/s and roughly 82K premium+JS requests. Real-world blended cost depends heavily on which tier your target domains require: 6 of our 20 benchmark domains worked on standard, 5 on standard+JS, 6 on premium, and 3 only on premium+JS.

Credits don't map one-to-one to requests. Each feature stacks a multiplier on top of the base call:

• ·Standard proxy (8 geos): $0.50 / 1K (1x base)
• ·Standard + JS render: $0.75 / 1K (1.5x)
• ·Premium proxy (195+ geos): $1.00 / 1K (2x)
• ·Premium + JS render: $1.50 / 1K (3x)
• ·AI Parser: Free — natural-language → JSON

A few things to watch out for before signing up:

• !Smartproxy → Decodo rebrand confusion still circulating in old docs.
• !Starter tier rate-capped at 10 req/s — must jump to Advanced for 25 req/s.
• !Target templates force the premium pool (2-3x cost).

Credit-based consumption rather than flat per-request, billing compute units, residential or SERP proxy traffic, storage I/O, and data transfer. Free tier includes $5 monthly credits; paid Starter, Scale, and Business tiers unlock more compute and discounted pay-per-event rates. Third-party Actors add rental fees from roughly $20/month or $8 per 1,000 results.

Credits don't map one-to-one to requests. Each feature stacks a multiplier on top of the base call:

• ·Compute units: CPU × memory × runtime per actor
• ·Residential proxy: Per-GB transfer
• ·SERP proxy: Per query
• ·Datacenter proxy: Per-GB (cheapest tier)
• ·Dataset / storage I/O: Per read + write

A few things to watch out for before signing up:

• !Compute-unit pricing across 10+ dimensions is hard to forecast.
• !Marketplace quality varies — community actors lack consistent maintenance.
• !Monthly actor rental and pay-per-event fees stack on top of platform compute.

Web Unlocker, SERP, and Crawl APIs start at $1.50 per 1,000 successful results on pay-as-you-go, dropping to $1.00 at the $1,999/month tier. The Scraping Browser is bandwidth-billed at $8/GB down to $5/GB, residential proxies start at $8/GB (currently $4 with promo), and datasets begin at $250 per 100K records. New accounts get $5 in free credits plus a first-deposit match up to $500.

Credits don't map one-to-one to requests. Each feature stacks a multiplier on top of the base call:

• ·Scraping Browser: $8/GB → $5/GB bandwidth
• ·Residential proxies: $8/GB ($4/GB promo)
• ·Datasets Marketplace: $250 per 100K records
• ·Archive API (historical): $1.00 / 1K vs $0.20 / 1K recent

A few things to watch out for before signing up:

• !Each product line bills separately even though tiers look unified.
• !Scraping Browser is GB-billed, not per request — costs vary with page weight.
• !Promo rates expire after the first cycle; renewals can effectively double.

Pricing starts with a free 1,000-credit tier, then a $30 Discovery plan with 200,000 credits, scaling to $500 Enterprise at 5.5M credits. Fully-protected requests using ASP, JS rendering, and residential proxies cost up to 30 credits each, so headline credit counts shrink quickly on hard targets. Annual billing knocks off roughly 16%.

Credits don't map one-to-one to requests. Each feature stacks a multiplier on top of the base call:

• ·Base request: 1 credit
• ·ASP (anti-bot bypass): +5 credits
• ·render_js (browser): +5 credits
• ·Residential proxy: +25 credits
• ·Fully-protected (ASP + JS + residential): ~30 credits

A few things to watch out for before signing up:

• !SMS phone verification at signup blocks some users from trialing the service.
• !$30 plan yields only ~6,600 fully-protected requests per month.
• !Annual billing saves ~16% but locks you in.

Web Scraper API starts at $49/month for 98K Amazon results, with success-based billing on 2xx/4xx responses only. Web Unblocker uses GB-based pricing from $9.40/GB regular ($75 entry plan), while residential proxies start at $6/GB and datacenter from $0.59/GB. Enterprise tiers require custom quotes.

Credits don't map one-to-one to requests. Each feature stacks a multiplier on top of the base call:

• ·Web Unblocker bandwidth: $9.40/GB ($5.64 promo)
• ·Residential Proxies: $6.00/GB
• ·Datacenter Proxies (PAYG): $0.59/GB
• ·Web Scraper API + JS rendering: $1.35 / 1K results

A few things to watch out for before signing up:

• !GB-based billing on Web Unblocker makes costs unpredictable vs flat per-request.
• !$75/mo Web Unblocker entry is the highest in the unblocking category.
• !No MCP server — integration requires SDKs, Postman, or direct HTTP.

Pricing is credit-based with a 500-credit free tier and a $17/month Starter plan that includes 10,000 credits and 100 requests per minute. Markdownify scrapes cost 1 credit, Extract calls 5 credits, and stealth mode adds 5 credits per request; top-up packs start at $5 for 1,000 credits.

Credits don't map one-to-one to requests. Each feature stacks a multiplier on top of the base call:

• ·Scrape (markdown): 1 credit
• ·Scrape (screenshot): 2 credits
• ·Extract (LLM-structured): 5 credits
• ·Search (with prompt): 5 credits / result
• ·Stealth toggle: +5 credits per request

A few things to watch out for before signing up:

• !Extract costs 5x a basic markdown scrape — the LLM is the expensive part.
• !Stealth adds a flat +5 credits, making protected targets ~6x more expensive.
• !Some hard sites (Zillow-class anti-bot) return empty bodies even with stealth.

Paid Scraper API tiers run from $19/mo for 7,000 calls up to $599/mo for 450,000, with a free 5,000-request trial. SERP and Amazon APIs price separately, starting at $28 and $59 respectively, and residential proxies begin at $8.50/GB pay-as-you-go.

Credits don't map one-to-one to requests. Each feature stacks a multiplier on top of the base call:

• ·Scraper API (success-billed): 1 credit
• ·Browser API (billed on failure too): 1 credit / request
• ·Heavier protected targets: Up to 5 credits
• ·SERP API: Separate $28/mo plan, 10K requests
• ·Amazon API: Separate $59/mo plan, 10K requests
• ·Residential Proxies PAYG: $8.50/GB

A few things to watch out for before signing up:

• !Starter plan ships just 7K requests — rivals offer 100K+ at the same price.
• !Browser API bills 100% of requests, including failures.
• !Naming clashes with ScraperAPI, ScrapingAPI, and ScrapingBee.

Paid plans run $49/mo Freelance (250k credits), $99 Startup (1M), $249 Business (3M), and $599 Business+ (8M), all excluding VAT. A 1,000-credit trial is offered with no card required, but there is no ongoing free tier.

Credits don't map one-to-one to requests. Each feature stacks a multiplier on top of the base call:

• ·JS rendering (default on): 5 credits
• ·Premium proxy without JS: 10 credits
• ·Premium proxy with JS: 25 credits
• ·Stealth proxy (JS required): 75 credits
• ·Google SERP (custom_google): 15 credits

A few things to watch out for before signing up:

• !JS rendering is on by default — 5 credits per call.
• !Stealth proxy costs 75 credits each and requires JS.
• !No ongoing free tier; only a one-time 1K-credit trial.

Zyte API uses a five-tier per-website model from $0.13 per 1,000 simple HTTP requests up to $16.08 for advanced browser-rendered ones, with tier assigned automatically by the platform. Volume commitments from $100 to $500 monthly unlock discounts of 25-52%, and a $5 trial credit lasts 30 days.

Credits don't map one-to-one to requests. Each feature stacks a multiplier on top of the base call:

• ·T1 Simple HTTP: $0.13 / 1K
• ·T3 Moderate HTTP: $0.44 / 1K
• ·T5 Advanced HTTP: $1.27 / 1K
• ·T1 Simple Browser: $1.01 / 1K
• ·T5 Advanced Browser: $16.08 / 1K (124x T1 HTTP)
• ·AI extraction (product/article): $0.0004 – $0.0016 each

A few things to watch out for before signing up:

• !Auto-tier means you can't predict cost — Zyte assigns each website T1-T5 with no override.
• !Cross-modal spread (T1 HTTP → T5 browser) is 124x.
• !Tiers reviewed quarterly with only 2 weeks notice.

Plans run from $69.99/month Developer (250K basic requests, 20 concurrency) through $2,999.99 Business 3K, plus custom Enterprise. Credits multiply 5x for JS rendering, 10x for premium proxy, and 25x combined. Top-ups add 15% usage for 15% price, and yearly billing saves 10%.

Credits don't map one-to-one to requests. Each feature stacks a multiplier on top of the base call:

• ·Basic request: 1x credits
• ·JS rendering only: 5x
• ·Premium proxy only: 10x
• ·JS rendering + premium proxy: 25x

A few things to watch out for before signing up:

• !No pay-as-you-go option — only monthly subscriptions.
• !Developer plan caps premium+JS at just 10K requests per month.
• !Geo-targeting is country-level only; no city or state targeting.

Pricing spans 28 monthly tiers from $40 LITE (200K credits, 5 concurrency) up to $30,000 NOVA PRO (1.1B credits). Annual billing trims roughly 17 percent, and only successful 200/404 responses are charged. A 1,000-credit free trial requires no card, but the free plan hard-caps at 1,000 requests per month.

Credits don't map one-to-one to requests. Each feature stacks a multiplier on top of the base call:

• ·Generic /scrape (base, no JS): 1 credit
• ·Generic /scrape + JS render (default!): 5 credits
• ·Generic /scrape + premium proxy: 10 credits
• ·Generic /scrape + premium + JS: 25 credits
• ·Dedicated endpoints (Amazon, Indeed, Yelp): 1 credit
• ·LinkedIn profile (protected): 100 credits

A few things to watch out for before signing up:

• !JS rendering is ON by default — silently costs 5x base credits.
• !LinkedIn profiles and Amazon Reviews cost 100 credits each.
• !Free plan hard-caps at 1K credits/month with a 403 on overage.

Paid plans start at $49/month Hobby (100,000 credits, 20 threads, US/EU geo only) and scale through Startup ($149), Business ($299, full country geo), and Scaling ($475). A 7-day trial offers 5,000 credits without a card, and a 1,000-credit/month permanent free tier remains available.

Credits don't map one-to-one to requests. Each feature stacks a multiplier on top of the base call:

• ·Standard request: 1x credits
• ·Premium proxy: 10x
• ·Premium + render: 25x
• ·Ultra premium: 30x
• ·Ultra premium + render: 75x

A few things to watch out for before signing up:

• !x.com / twitter.com on ToS denylist — blocked at API level on every tier.
• !Hobby and Startup restricted to US/EU geo; country geo unlocks at $299.
• !75x credit range forces per-domain budgeting to avoid silent quota burn.

Subscription-only tiers run from a free 500-credit trial through Hobby ($16/mo, 3K credits), Standard ($83/mo, 100K), Growth ($333/mo, 500K), and Scale ($599/mo, 1M). There is no pay-as-you-go option, plan credits never roll over, and JSON or enhanced-proxy modifiers stack additional credits per page.

Credits don't map one-to-one to requests. Each feature stacks a multiplier on top of the base call:

• ·/v1/scrape (basic): 1 credit / page
• ·/v1/search: 2 credits per 10 results
• ·Browser sessions / interact: 2 credits / browser-minute
• ·JSON (LLM extraction): +4 credits / page
• ·Enhanced (stealth) proxy: +4 credits / page
• ·FIRE-1 agent: Dynamic; billed on failures too

A few things to watch out for before signing up:

• !No pay-as-you-go — subscriptions only, and credits never roll over.
• !FIRE-1 agent bills even on failed requests, rate-limited to 10/min.
• !Enhanced proxy is US + Denmark only; auto-recharge capped at 4 packs/mo.

Plans run from a free 10,000-credit tier to Enthusiast at $19 for 100K credits, Startup at $49 for 500K, and Business at $249 for 3M. Credit multipliers matter: residential plus JS rendering burns 125 credits per call, so a nominal 100K plan can deliver as few as 800 protected requests.

Credits don't map one-to-one to requests. Each feature stacks a multiplier on top of the base call:

• ·Datacenter proxy: 1 credit
• ·JS render (browser + datacenter): 10 credits
• ·Residential proxy: 25 credits
• ·Residential + JS render: 125 credits

A few things to watch out for before signing up:

• !browser=true is default — JS render silently costs 10x credits per call.
• !423 anti-bot errors surface on hard targets (Google SERP, fingerprinted sites).
• !No structured-data endpoints — relies on AI extractor for parsed output.

Pricing starts at $25/month for 1,000 searches on Starter, scaling through Developer ($75), Production ($150), and Big Data ($275 for 30,000 searches). Every successful call counts as one search regardless of engine, while cached, errored, and failed requests are not billed. Ludicrous Speed costs 2x per search.

Credits don't map one-to-one to requests. Each feature stacks a multiplier on top of the base call:

• ·Any engine (Google, Bing, Amazon, YouTube, etc.): 1 credit / search
• ·Ludicrous Speed (priority queue): 2x base price
• ·Cached repeat within 1h: Free (not counted)

A few things to watch out for before signing up:

• !SERP-only — no arbitrary URL scraping or product / profile pages.
• !$0.025 / search on Starter is ~10x the per-request rate of generalist scrapers.
• !Each engine returns a different response schema; integration scales with engine count.