ScrapingBee review

Pricing and credits

ScrapingBee sells monthly credit bundles (as of October 2026, per its pricing page). Every plan has the same features, so the choice is purely about volume and concurrency.

A credit is not a request. What a request costs depends on its options:

• Rendering is on by default. A bare request runs a headless browser for 5 credits. For server-rendered pages, send render_js=false: on Freelance that is $0.20 instead of $0.98 per 1,000.
• Failures are free, and auto mode (mode=auto) escalates and charges only for the configuration that worked, so easy pages don't pay stealth prices.
• Unused credits expire. Size for a normal month, not your peak. AI extraction adds 5 credits.

The 1,000 trial credits buy 200 JS requests or 40 premium + JS ones: test your hardest target first.

The API in practice

One endpoint, https://app.scrapingbee.com/api/v1, with options as query parameters. The docs prefer the key in an Authorization: Bearer header (the api_key parameter still works).

# Server-rendered page, no browser: 1 credit
curl -G "https://app.scrapingbee.com/api/v1" \
  -H "Authorization: Bearer $SCRAPINGBEE_API_KEY" \
  --data-urlencode "url=https://example.com/products?page=2" \
  -d render_js=false

# Residential (premium) IP in Germany, with JS rendering: 25 credits
curl -G "https://app.scrapingbee.com/api/v1" \
  -H "Authorization: Bearer $SCRAPINGBEE_API_KEY" \
  --data-urlencode "url=https://example.com/products?page=2" \
  -d premium_proxy=true -d country_code=de

Also available: stealth_proxy, LLM extraction (ai_query, ai_extract_rules), screenshots and file downloads.

Google search

The Google API (query in search) costs 10 credits as a light request, the default, or 15 for fuller results. search_type covers classic, news, maps, images, lens, shopping, ai_mode and ads. That is $1.96 per 1,000 searches on Freelance, $2.53 on Hobby. Fast Search returns Google results "under a second", by ScrapingBee's account, for 10 credits per successful call.

AI agents and MCP

The remote MCP server at https://mcp.scrapingbee.com/mcp takes the key as ?api_key=. Its 18 tools cover page text and HTML, extraction, screenshots, files, Google and Fast Search, ask_chatgpt, ask_gemini, Amazon, Walmart and YouTube lookups, and usage. Docs cover Claude Desktop, Cursor and ChatGPT; the first two connect through mcp-remote:

{
  "mcpServers": {
    "scrapingbee": {
      "command": "npx",
      "args": ["mcp-remote", "https://mcp.scrapingbee.com/mcp?api_key=YOUR_API_KEY"]
    }
  }
}

Calls cost the same credits as the API, with no surcharge. Treat the config file as a secret, since the key is in the URL, and add retries: the hosted server shares a global concurrency limit and can return HTTP 429 even when your plan has headroom. Compare it in our AI-agent search & fetch guide.

Independent evidence

Proxyway's Web Scraping API Report 2025 (tests October 2025, published 3 December 2025) ran ten APIs against 15 protected sites at 2 requests per second. ScrapingBee reached 84.47%, fourth, just behind Decodo (85.88%) and Oxylabs (85.82%) and well behind Zyte (93.14%). It was the slowest at 25.46 seconds per response, and managed 5% on G2.

ScrapeOps' fingerprint test (January 2026, higher is better) gave it 24.76, behind Zyte (80.48), Bright Data's Unlocker (41.43), Oxylabs (30.00) and ScraperAPI (27.81). ScrapeOps' Cloudflare playbook (June 2026) put scraping APIs as a group at about 97% on Cloudflare sites, a category figure rather than a ScrapingBee one. These are third-party tests, not neutral audits: Proxyway uses affiliate links and ScrapeOps runs a paid proxy aggregator.

Ownership: Oxylabs and Source.ai

Oxylabs' group acquired ScrapingBee (announced 19 June 2025), and it has kept running as its own product. On 13 October 2026 the group launched Source.ai, an umbrella over Oxylabs, Decodo, Webshare, ScrapingBee and Honeygain; the brands keep their own products. ScrapingBee, the Oxylabs Web API and Decodo now share a parent; for a fallback outside the group, look at Zyte, ZenRows or ScraperAPI.

Where it fits

Our guides make picks by use case. ScrapingBee is the pick for:

• Most complete search + fetch toolkit (Search & fetch APIs for AI agents guide)
• Every feature on every plan, with excellent docs (Web scraping APIs guide)

Why: Google search and JS-rendering fetch, screenshots and extraction sit behind one key and an 18-tool MCP server; 25,000 searches a month fit the $49 plan ($1.96 per 1,000); and every feature is on every plan from $19.

Other tools fit better when:

• Success rate is everything: Zyte API was first of ten in Proxyway's 2025 report (93.14%, 11.15 s average), against ScrapingBee's 84.47% and 25.46 s.
• You want a flat price on hard targets: Bright Data Web Unlocker charges $1.50 per 1,000 with rendering and CAPTCHA solving included (97.90% in Proxyway's 2024 report; it skipped 2025), against $4.90 for premium + JS on Freelance.
• Your agent searches lightly or in bursts: Bright Data's SERP API charges $1.50 per 1,000 successful requests pay-as-you-go with 5,000 free records a month, and PacketStream $2.50 per 1,000 successful searches from a balance that never expires; ScrapingBee's minimum is $19 a month.
• You want a free monthly allowance: ZenRows gives 5,000 credits a month with no card, against ScrapingBee's one-time 1,000.

See also ScraperAPI vs ScrapingBee, ZenRows vs ScrapingBee and how we pick.

Pros and cons