REVIEW
ScraperAPI review
One GET endpoint, credit multipliers for rendering and premium proxies, fixed rates for Amazon, Google and LinkedIn, and an MCP server. Cheap per page on easy sites; the bill climbs fast on hard ones.
Verdict
ScraperAPI is a straightforward way to fetch easy and medium-difficulty pages: one GET endpoint, a 7-day trial with 5,000 credits and no card, and only 200 and 404 responses billed. Cost and hard targets are its weak spots. The entry plan is $49/mo, rendering costs 10 credits, premium proxies plus rendering is $12.25 per 1,000 pages on that plan, and in Proxyway's October 2025 test it managed 0% on Hyatt (Kasada) and Lowe's (Akamai) for a 68.95% average, sixth of ten. It was also quick (13.92 s average, about half ScrapingBee's time) and reached 99.97% on G2.
Best for
- Targets it handles well, such as G2 (99.97% in Proxyway's 2025 test).
- Amazon, Walmart, eBay or Google at fixed per-domain credit rates.
- Jobs where speed matters: 13.92 s average in that test, against ScrapingBee's 25.46 s.
Skip if
- Your key targets sit behind Kasada or Akamai (0% on Hyatt and Lowe's in 2025).
- Most requests need premium proxies plus rendering ($12.25 per 1,000 on Hobby).
- Your agent runs lots of Google searches (25 credits each).
- You need geotargeting outside the US and EU on a small budget.
- Entry plan
- Hobby $49/mo: 100,000 credits, 20 threads
- Trial
- 7 days, 5,000 credits, no card
- Credits per request
- 1 normal, 10 render or premium, 25 both, 75 ultra-premium + render
- What is billed
- 200 and 404 responses, plus requests cancelled before 70 s
- Support
- Email, weekdays, below enterprise (Proxyway)
Pricing and plans
ScraperAPI sells monthly credit subscriptions (as of October 2026, per its pricing page), 10% cheaper billed annually.
| Plan | Per month | Credits | Threads | Geotargeting | Per 1,000 credits |
|---|---|---|---|---|---|
| Hobby | $49 | 100,000 | 20 | US & EU | $0.49 |
| Startup | $149 | 1,000,000 | 50 | US & EU | $0.15 |
| Business | $299 | 3,000,000 | 100 | Country-level | $0.10 |
| Scaling | $475 | 5,000,000 | 200 | Country-level | $0.10 |
| Professional | $975 | 10,500,000 | 300 | Country-level | $0.09 |
| Advanced | $1,975 | 21,500,000 | 500 | Country-level | $0.09 |
The step that matters is Hobby to Startup: three times the price for ten times the credits, so 1,000 credits fall from $0.49 to $0.15. Hobby's 100,000 credits buy 100,000 plain requests, 10,000 rendered ones or just 4,000 premium + render requests. If your targets need premium proxies, price the job on Startup.
How credits add up
| Request type | Credits | Hobby | Startup | Business |
|---|---|---|---|---|
| Normal request | 1 | $0.49 | $0.15 | $0.10 |
| JS rendering render=true | 10 | $4.90 | $1.49 | $1.00 |
| Premium (residential) proxies premium=true | 10 | $4.90 | $1.49 | $1.00 |
| Premium + rendering premium=true&render=true | 25 | $12.25 | $3.73 | $2.49 |
| Ultra-premium (paid plans only) ultra_premium=true | 30 | $14.70 | $4.47 | $2.99 |
| Ultra-premium + rendering ultra_premium=true&render=true | 75 | $36.75 | $11.18 | $7.48 |
| Amazon, Walmart, eBay | 5 | $2.45 | $0.75 | $0.50 |
| Google, Bing | 25 | $12.25 | $3.73 | $2.49 |
| 30 | $14.70 | $4.47 | $2.99 |
Anti-bot bypasses (Cloudflare, DataDome, Akamai and others) add 10 credits; country_code,
session_number, device_type, output_format and autoparse are
free. Two billing rules belong in your client code:
- Only 200 and 404 responses are charged. Other failures are free, but a crawl full of dead links pays for every 404.
- Requests you cancel before 70 seconds are charged too. ScraperAPI recommends a 70-second client timeout; a 30-second default means paying for responses you never read.
The API in practice
Everything goes through https://api.scraperapi.com with api_key, an encoded
url and optional switches:
# Normal request: 1 credit
curl --max-time 70 -G "https://api.scraperapi.com/" \
--data-urlencode "api_key=$SCRAPERAPI_KEY" \
--data-urlencode "url=https://example.com/products?page=2"
# Rendered through premium (residential) proxies: 25 credits
curl --max-time 70 -G "https://api.scraperapi.com/" \
--data-urlencode "api_key=$SCRAPERAPI_KEY" \
--data-urlencode "url=https://example.com/products?page=2" \
-d render=true -d premium=true
Escalate deliberately: normal first, render=true only for pages that need JavaScript,
ultra_premium only for domains that fail everything else. Proxyway found the top configurations of
credit-based APIs costing up to 100 times their base rate.
MCP server for AI agents
ScraperAPI documents a remote MCP server at https://mcp.scraperapi.com/mcp and publishes an MIT-licensed
local server (pip install scraperapi-mcp-server, or Docker) that reads the key from
API_KEY. Its scrape tool takes the API's switches (render,
premium, ultra_premium, country_code and more), alongside structured-data tools
for Google, Amazon, Walmart, eBay and Redfin, plus crawler and AI-parser tools. The local config from the project's
README:
{
"mcpServers": {
"ScraperAPI": {
"command": "python",
"args": ["-m", "scraperapi_mcp_server"],
"env": { "API_KEY": "<YOUR_SCRAPERAPI_API_KEY>" }
}
}
} We could not confirm how MCP calls are billed. At the API's rates, a Google search (25 credits) is $12.25 per 1,000 on Hobby, several times what ScrapingBee or PacketStream charge (see the our AI-agent search & fetch guide).
Independent evidence
In Proxyway's Web Scraping API Report 2025 (tests October 2025, published 3 December 2025; 15 protected sites at 2 requests per second) ScraperAPI averaged 68.95%, sixth of ten. The average hides a split:
- 99.97% on G2, where ScrapingBee managed 5% and Oxylabs 0%.
- 0% on Hyatt and Lowe's, protected by Kasada and Akamai respectively.
- A 13.92-second average, against 25.46 s for ScrapingBee and 11.15 s for Zyte, first in that test.
ScrapeOps' fingerprint test (January 2026, higher is better) gave it 27.81, near ScrapingBee (24.76) and far behind Zyte (80.48). Its Cloudflare playbook (June 2026) put scraping APIs as a group at about 97% on Cloudflare sites, a category figure rather than a ScraperAPI one. Caveats: Proxyway uses affiliate links and ScrapeOps runs a paid proxy aggregator. Trustpilot shows 4.5 from only 42 reviews. If your list includes Kasada or Akamai sites, spend the trial credits on those first.
Where it fits
Our guides make picks by use case. ScraperAPI has no pick in them; it appears in the comparison tables.
- Strongest independent evidence: Zyte API, first of ten in Proxyway's 2025 report at 93.14%, against ScraperAPI's 68.95% with 0% on Hyatt and Lowe's.
- Flat price with rendering included: Bright Data Web Unlocker at $1.50 per 1,000 (97.90% in Proxyway's 2024 report; it skipped 2025), against $12.25 for premium + render on Hobby.
- Every feature on every plan: ScrapingBee, with premium + JS at $4.90 per 1,000 on its $49 plan and 84.47% in Proxyway's 2025 benchmark.
- Search for AI agents: ScrapingBee ($1.96 per 1,000 on its $49 plan) or PacketStream ($2.50 per 1,000, no subscription), against $12.25 per 1,000 Google searches here.
Where ScraperAPI still fits: sites it handles well, such as G2 (99.97% in Proxyway's 2025 test, where ScrapingBee managed 5% and Oxylabs 0%); jobs where speed matters (13.92 s average against ScrapingBee's 25.46 s); and Amazon, Walmart, eBay or Google work, where fixed per-domain credit rates keep bills predictable. Compare it directly in ScraperAPI vs ScrapingBee, try your own volumes in the cost calculator, and see how picks are chosen on the methodology page.
Pros and cons
Pros
- One GET endpoint: api_key and url, then switches such as render=true.
- Only 200 and 404 responses billed; geotargeting and sessions cost nothing extra.
- 99.97% on G2 and a 13.92 s average in Proxyway's 2025 test.
- Fixed per-domain rates for Amazon, Walmart, eBay and Google.
- Remote MCP server plus an MIT-licensed local one.
Cons
- 0% on Hyatt (Kasada) and Lowe's (Akamai); 68.95% overall, sixth of ten.
- Premium + render is $12.25 per 1,000 on Hobby; ultra-premium + render $36.75.
- Rendering costs 10 credits, twice ScrapingBee's and ZenRows' 5.
- Requests you cancel before 70 seconds are billed.
- Hobby and Startup limited to US and EU geotargeting; email-only weekday support.
Alternatives
- ScrapingBee : higher 2025 success and cheaper credits at every price point, but slower.
- ZenRows : a free 5,000-credit monthly tier and a 25-credit ceiling per request.
- Zyte API : first of ten in Proxyway 2025 (93.14%), billed per site tier.
- Bright Data Web Unlocker : $1.50 per 1,000 successful requests, rendering included.
- Crawlbase : pay-as-you-go instead of a subscription.
- ScraperAPI vs ScrapingBee
- Scraping APIs for protected sites : our scraping API guide.
Sources
Vendor pages checked 2026-10-17. Prices are as of October 2026 and change often; check the vendor's page before buying.