Pricing guide · provider details reviewed September 2026

Web scraping API pricing: compare cost per successful result

A workload-first framework for choosing a provider

A low nominal price can be misleading when providers meter different things. One service may count a raw HTML request, another a browser-rendered credit, another compute or data transfer, and another a normalized record. Compare the cost of a successful, validated observation for your own request shape.

Freshness note: prices, credits, limits, and features change. The provider links below were reviewed on September 5, 2026. Use their official pricing and documentation pages before purchase; this article deliberately avoids an unmaintainable price snapshot.

Pricing models at a glance

Service	What to compare	Official source
ExtractAPI	Monthly request allowance for a bounded REST extraction. Current app configuration: 50 trial requests, 500/month after verification, Starter 5,000/month, Pro 25,000/month, and Business contact-only.	Plans · API fields
ScrapingBee	Monthly API credits. Its docs explain that JavaScript rendering and proxy choices can change credit cost.	Pricing · HTML API docs
ScraperAPI	Credits plus request options. Its docs describe additional cost for JavaScript rendering and premium proxy modes.	Pricing · Rendering cost docs
Scrapingdog	Credits and concurrency by plan; its documentation lists separate costs for rotating proxies, JavaScript rendering, and premium proxy options.	Pricing · Request customization
Apify	Platform usage such as actor compute, data transfer, proxy use, and storage operations rather than a universal request count.	Usage docs
Bright Data	Product-specific records or usage, concurrency, and delivery terms; confirm the exact Web Scraper API offering.	Web Scraper API pricing

Use one denominator

For a representative period, record:

valid target observations requested and validated;
rendering, proxy, geography, screenshot, or extraction-rule options used;
failed, blocked, timed-out, retried, and duplicate responses;
response parsing, storage, queue, and support costs you operate yourself.

Then calculate (provider charge + your operating cost) / validated observations. Keep an “unknown” bucket when a response is incomplete; treating every HTTP 200 as a successful extraction hides the real cost.

Why browser rendering changes the estimate

JavaScript rendering can reveal content that is absent from the initial HTML, but it consumes more resources and may have a different provider price. Test a raw request and a browser request against the same permitted target, then measure whether the extra output is worth the cost.

curl -X POST https://extractapi.app/v1/extract \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"url":"https://demo.example/products/widget","javascript":true}'

ExtractAPI's response contains structured fields under data, including mainContent and structuredData. A selected CSS value is returned in data.custom. See the JavaScript website to JSON workflow for a synthetic response and limitations.

Read the current terms, not an old table

Before committing, verify billing interval, trial terms, credit expiry, failed-request treatment, concurrency, regional behavior, data retention, support, and cancellation. A provider’s pricing page and request documentation are the source of truth. Keep a dated internal record when the workload is material.

For ExtractAPI, the local source of truth is the runtime plan response backed by stripe.js. The Business tier is contact-only while fixed volume and service commitments remain unverified. No annual discount, overage promise, or SLA is implied by this guide.

Choose by workload

Small structured extraction: compare the complete JSON shape, selector behavior, and fixed allowance against your validation needs.
Large or variable workloads: model concurrency, retries, proxy or geography options, and usage-based charges with a real sample.
Client-rendered pages: benchmark the required field with JavaScript on and off, and keep a fallback for missing or blocked output.
Protected targets: obtain permission or use an approved feed. Do not choose a provider on the assumption that it can bypass controls.

FAQ

Which scraping API is the cheapest?

There is no universal winner. The answer depends on successful output, request options, volume, retries, concurrency, and your own operating cost. Use the denominator above and current official pricing.

Are credits the same as requests?

Not always. Rendering, proxy, geography, AI, or site-specific modes can consume multiple credits. Read the provider’s cost rules for the exact endpoint and parameters you will use.

Does a free tier prove production readiness?

No. Validate target coverage, limits, failure handling, data quality, support, and terms with representative permitted targets before relying on a service.

Get a free API key