Apify Review

Apify is two products under one name. The first is the Apify Store, a marketplace of ready-made scrapers called Actors, many of them written by outside developers. The second is the cloud that runs those Actors, with proxies, storage and an API. If you need a scraper for a site and do not want to maintain its selectors, that is the product to price. If you need raw proxy bandwidth, you are buying the wrong thing.

The real question is what the bill looks like once your runs are real. Toolradar data: Toolradar tracks 1,201 user reviews of Apify on G2 and Capterra, averaging 4.7 as of 10 October 2026. Apify's pricing page, checked on 10 October 2026, sets four plans and three compute-unit rates, so the cost depends on how many units your Actors burn. For the wider field, start with the scraping shortlist; Toolradar's own Apify listing has the tool details. This review works through that billing, prices four alternatives on their own vendor pages, and says where Apify loses.

Apify is a web scraping and automation platform built around Actors: small programs that do one job, such as pulling listings from a marketplace, crawling a set of pages or collecting search results. You can run a ready-made Actor from the Store, or write your own in JavaScript, TypeScript or Python and publish it.

Under the Actors sits the platform itself, with cloud compute, residential and datacenter proxies, dataset and key-value storage, and an API. Apify is what you buy when you want the scraper and the infrastructure from one vendor.

The platform is also open to AI agents. An MCP server gives an assistant access to Actors, and the pricing page says an AI agent can pay for an Actor run without an Apify account.

That makes the billing model matter more, because an agent will run an Actor without anyone watching the meter. If the scraped data feeds a model, the LLM API price comparison shows what the tokens cost.

You pick an Actor from the Store, give it an input such as a URL, a search term or a list of IDs, run it, and read the dataset it writes. Your Billing settings show your platform usage, which is where you find out what an Actor really costs.

The metering is where Apify differs from a flat subscription. Every run is billed in compute units. One compute unit is one gigabyte of memory for one hour, so an Actor that holds 4 GB for an hour uses 4 CU. That is $0.80 on Starter, $0.64 on Scale and $0.52 on Business.

Proxies, dataset storage and data transfer draw on their own meters during the same run. On Free and Starter, dataset timed storage costs $1.00 per 1,000 GB-hours and external data transfer costs $0.20 per GB.

Each paid plan comes with prepaid platform usage equal to its monthly fee. Under that amount you pay the fee and nothing more. Above it, the excess goes on the next invoice, so your bill is the higher of the plan fee and your usage at that plan's rates.

Unused credit expires at the end of the billing cycle and does not roll over. Actors in the Store use one of two models: pay per event, where the developer sets a fixed price for each event, or pay per usage, where you pay only for the platform usage the run consumes.

Check the model on each Actor's page before you schedule it. Keeping the results in your own database is a separate cost, and the free database hosting guide covers the options.

Apify publishes four plans on one page. The Apify figures come only from that page, checked on 10 October 2026. On monthly billing, Free is $0 with $5 of usage, and the three paid plans are listed in the table below. Paid plans take 10% off when billed annually. On Free, access stops once the allowance is used, until the next monthly cycle.

To see what a bill looks like, we costed one workload: 400 compute units and 5 GB of residential proxy in a month, the sort of load a small team runs for a daily price monitor. The units and gigabytes are our assumption, not a vendor figure. The arithmetic uses only the rates on the page.

Starter: 400 x $0.20 = $80 of compute, plus 5 x $8 = $40 of proxy, so $120 of usage. The bill is the higher of the plan fee and $120, which is $120 a month.

Scale: 400 x $0.16 = $64, plus 5 x $7.50 = $37.50, so $101.50 of usage. The bill is the plan minimum.

Business: 400 x $0.13 = $52, plus 5 x $7 = $35, so $87 of usage. The bill is the $999 minimum.

Free: its allowance covers 25 compute units at $0.20, so this workload cannot run on Free.

Two crossovers matter. Scale costs less than Starter once usage at Starter rates passes about $200 a month, which is Scale's $199 minimum. Past that point its compute rate, 20 percent lower, widens the gap. Business only beats Scale once usage at Scale rates passes about $1,000. Below those points, a higher plan costs you its minimum for no saving. A team that would rather run the same crawl on its own server should price a cheap VPS against the Apify bill.

Methodology. Apify prices come from Apify's pricing page, checked on 10 October 2026. Alternative prices come from Firecrawl pricing, Zyte pricing, Bright Data's Web Unlocker page, Bright Data's proxy page and Oxylabs pricing, checked the same day. The review count comes from Toolradar's Apify listing, read on 10 October 2026. Toolradar tracks 11,051 tools across 343 categories. The scenario leaves out annual billing and the Apify Store discounts.

Apify fits a developer or small data team that needs scrapers running on a schedule and does not want to build or host them. It makes sense when an Actor already covers your target sites, such as marketplaces, maps, search results or job boards, because then you pay for runs, not for development. If the job is one scrape a month, the Free plan covers it, and the Crawlee library gets you there without a subscription.

It is the wrong tool for raw proxy volume. At 1,000 GB or more, Oxylabs residential falls to $2.50 per GB, while Apify's residential rate stays higher even on Business. It also suits a team poorly when usage swings: a quiet month costs the plan fee, and a busy month can cost several times that. Teams that mainly need page content for an AI pipeline should compare Firecrawl, whose plans are priced per credit, and its Toolradar tool page gives the product overview. Proxy-heavy jobs belong in the residential proxy guide. If the job is a workflow with a scraper inside it, the n8n review covers that model, and the serverless platforms guide covers running your own code.