Apify review

Pricing

Apify sells prepaid usage, not pages. As of October 2026 (per the Apify pricing page) Free includes $5 of usage a month, and Starter ($19/mo), Scale ($199) and Business ($999) each include their price in usage, 10% cheaper billed yearly. Overage on a paid plan goes on the next invoice; on Free, access stops until the next cycle. Unused usage does not roll over, and the free plan's proxy works only inside Apify.

What a job actually costs

A compute unit (CU) is 1 GB of RAM for one hour, so the free $5 buys 25 CU, or about six hours of a 4 GB browser Actor. Apify's own advice is to do a test run and read the usage. A worked example, with assumed throughput:

• Compute: a 4 GB Playwright Actor doing 1,000 pages in 30 minutes uses 2 CU: $0.40.
• Bandwidth: about 500 KB a page with assets blocked is 0.5 GB per 1,000 pages: $4.00 at $8/GB ($16 for 2 MB full renders).
• Total: about $4.40 per 1,000 pages, roughly 90% of it proxy traffic.

So on browser jobs the proxy rate matters most: $8/GB is twice the $4.00/GB median entry price in Proxyway's 2026 market research and eight times PacketStream's $1/GB, which you can use with Crawlee on your own servers. Google results are cheaper to budget: $2.50 per 1,000 SERPs on Starter, $1.70 on Business. Try your own numbers in the web scraping cost calculator.

Actors and the Store

An Actor is a packaged scraper or automation that runs on Apify's cloud. The Store lists 19,000+ of them, and developers get paid for publishing Actors, so popular sites often have several competing scrapers of uneven quality. Each Actor's page states its pricing model: pay per event (a fixed price for events such as a start or a returned result, with platform usage included in most cases) or pay per usage (no developer fee; you pay for the compute, storage, data transfer and proxies the run consumes). Per-event pricing is easier to budget; per-usage Actors are only as cheap as their code is efficient.

Crawlee and self-hosting

The open-source part is Crawlee, Apify's Apache-2.0 crawling library for JavaScript/TypeScript and Python (v3.18.2): HTTP and browser crawlers, a session pool, proxy rotation through ProxyConfiguration and blockRequests() to trim asset downloads in Chromium. The platform itself is a hosted service, but Crawlee code runs on any server with any proxy:

import { PlaywrightCrawler, ProxyConfiguration, Dataset } from 'crawlee';

const crawler = new PlaywrightCrawler({
  proxyConfiguration: new ProxyConfiguration({ proxyUrls: [process.env.PROXY_URL] }),
  async requestHandler({ request, page, enqueueLinks }) {
    await Dataset.pushData({ url: request.loadedUrl, title: await page.title() });
    await enqueueLinks();
  },
});

await crawler.run(['https://example.com/']);

A sensible path: prototype with a Store Actor, and if the bill or success rate disappoints, move the job to Crawlee on your own infrastructure. Crawlee is our pick for browser crawling with proxies and sessions built in in the open-source guide; the build vs buy guide covers when moving off a hosted platform pays off.

Independent evidence

Proxyway's Web Scraping API Report 2025 (tests October 2025, published 3 December 2025) tested Apify separately from the ten APIs in its main table. Dedicated Actors "failed to work with around half of the targets"; a Walmart run took 14 hours and a Leboncoin job cost more than $60 for a tenth of the results, while the G2 Actor performed excellently. Proxyway uses affiliate links, the test is a year old and Actors change as their developers update them, so read it as a warning about variance, not a verdict on every Actor. Trustpilot is warmer, at 4.8/5 from 802 reviews, but it reflects whatever reviewers use Apify for rather than protected targets.

Where it fits

Our guides pick products by use case (how we pick). In the no-code guide Apify is the pick for ready-made scrapers for popular sites: no other tool we cover has a marketplace of 19,000+ Actors with schedules, webhooks, an API and built-in proxies.

Other tools fit other needs better:

• Building your own scraper without code: Octoparse is a point-and-click desktop builder with cloud runs, so you are not limited to what the Store covers.
• Protected targets, paying only for success: scraping APIs such as Zyte API (93.14% in Proxyway's 2025 benchmark) or Bright Data's Web Unlocker ($1.50 per 1,000) bill per successful page rather than per compute unit and GB. See the scraping API guide.
• High-volume browser jobs: residential traffic at $7–$8/GB dominates the bill; running Crawlee yourself with cheaper bandwidth from our proxy guide is the usual way out.

Pros and cons