Instead of running your own client through a proxy, you call an HTTP endpoint with the target URL as a parameter, and the service returns the page. Behind that call it picks a proxy, sets browser-like headers and fingerprints, retries failures, and often renders JavaScript in a headless browser. Some also parse common page types, such as search results or product pages, into JSON. A variant called an unblocker works as a proxy endpoint you point your existing client at.
How they are priced
Usually per request, sometimes per successful request, rather than per gigabyte. Many multiply the cost of a request for options such as JavaScript rendering or premium proxy types, so the price on the page can be several times lower than the price of the requests you make. Work out the cost per 1,000 pages you need, with the options you will use, and compare it with proxies at your measured page weight.
When an API fits
- A few hard targets that need rendering and challenge handling you do not want to build.
- Small or irregular volume, where engineering time costs more than the requests.
- Teams without scraping infrastructure.
When proxies fit better
- High volume on moderate targets, where per-request pricing adds up fast.
- Sessions and logins, which need one identity across many requests.
- Full control over the browser, the timing and what gets downloaded.
Many teams use both: proxies with their own client for most of the work, an API for the targets that resist.
Common confusion
A scraping API does not change what you are allowed to collect; the site's terms and the law apply the same way. And "success" in an API's pricing is defined by the vendor, so check whether a CAPTCHA page returned with a 200 counts. Success rate covers how to measure it yourself.