Apify Pricing in 2026: Compute Units Explained (Real Workload Math)

A team starts an Apify trial on the free $5 credit, ships a working prototype in two days, moves to the $29 Starter plan, and three months later their finance dashboard shows the line item at $387 a month. Nobody on the team can quickly say why.

That story is not unusual, and it is not Apify's fault. Apify's platform is genuinely powerful. Its actor marketplace, browser farm, and proxy infrastructure handle problems most teams cannot afford to build themselves. The challenge is that the bill is the sum of five separate variables, and the platform's headline price is only the prepaid floor. This guide is the explainer that decodes the whole picture: every tier, the Compute Unit math, what a typical workload actually costs, and where the dial points when your prototype becomes production.

Every Apify number below was checked directly against apify.com/pricing on May 23, 2026. Pricing pages move, so click through and verify the live numbers if you are reading this in a later quarter. At the end, there is an honest comparison with an alternative API that prices on pages, not Compute Units, for teams that want predictable monthly bills.

The short answer: Apify pricing at a glance

Apify offers four published tiers and a custom Enterprise option. Each tier sets a monthly prepaid platform credit that is consumed by your actors at a Compute Unit rate. The cheapest CU rate sits on the largest plan.

Source: apify.com/pricing, verified 2026-05-23.

That is the rate card. Now the part the pricing page does not explain: what a Compute Unit actually is, and how it translates into pages crawled, runs completed, and dollars spent.

What is a Compute Unit on Apify?

A Compute Unit on Apify equals 1 GB of actor memory running for 1 hour. So a 4 GB actor running for 15 minutes consumes 1 CU. A 1 GB actor running for 4 hours also consumes 1 CU. The unit is a memory-time product, which means the same job can cost dramatically different amounts depending on how the actor is configured and how long each page takes.

This matters because actor RAM is decided by the actor author, not by you. A scraper from the Apify Store might be set to 4 GB by default. If the run lasts 30 minutes, you have spent 2 CU, or $0.40 on the Starter rate. Run it 50 times in a day and you have spent $20 of your $29 monthly credit before lunch.

Two practical rules to keep in mind:

1. RAM caps multiply the bill. The same page processed by an 8 GB actor costs twice the CU of a 4 GB actor. Always check the default RAM on any actor you import from the store. The actor's developer optimized for their convenience, not yours.
2. The CU rate drops as the plan goes up. $0.20 on Free and Starter, $0.16 on Scale, $0.13 on Business. The $999 Business plan is paying for the cheaper unit rate as much as the higher credit allowance. For high-volume teams, the unit-rate discount is what makes the upgrade pay back.

One more property worth knowing: pay-as-you-go (PAYG) is on by default once you cross the prepaid credit. Crossing the credit ceiling does not pause your pipeline; it bills the next CU at the same rate against your card. That is correct behavior for production teams, and it is a surprise for prototype teams who expected a hard cap.

Walking through each Apify tier

Free: $5 credit per month, $0.20 per CU

The Free tier exists for prototyping. $5 of credit at $0.20 per CU is 25 Compute Units, which is enough to run a 4 GB actor for about 6 hours total in a month. That comfortably covers wiring Apify into a side project, running a hackathon weekend, or evaluating an actor from the store before committing to a plan.

The Free tier is not a production option. A daily run of even a small scrape lasting 20 minutes on a 4 GB actor consumes 20 CU per month, plus weekend exploration, plus the next thing you try. The $5 credit is gone in days. Treat it as evaluation, not deployment.

Starter: $29 a month plus pay-as-you-go

Starter is the entry paid tier. You get $29 of platform credit (145 CU at $0.20 per CU), 25 concurrent runs, 32 GB max actor RAM, and 30 included datacenter proxy IPs. Anything above the credit bills as overage at the same rate.

For a single developer running a daily small-to-medium scrape, $29 a month is workable, but only if you watch the actor RAM. A 4 GB actor running a 30-minute job once a day consumes 60 CU a month, which is already $12. Two such jobs and you are at $24, and you have not bought any proxy IP overage yet. The plan covers a single solid use case; it does not cover three.

Scale: $199 a month plus pay-as-you-go

Scale is where production teams live on Apify. $199 of credit at the $0.16 per CU rate gives you roughly 1,244 CU per month. The concurrency cap jumps to 128, the RAM cap to 128 GB, and 200 datacenter IPs are included with $0.80 per extra IP. This tier supports sustained workloads: a daily multi-thousand-page scrape, weekly site refreshes, or continuous monitoring across a fleet of actors.

The unit-rate drop from $0.20 to $0.16 (a 20% improvement) is what makes Scale attractive over running Starter with high overage. If your monthly compute is heading north of about 1,000 CU, Scale becomes cheaper than Starter PAYG by simple arithmetic. Most teams cross that line within six months of going live.

Business: $999 a month plus pay-as-you-go

Business is the highest published tier before custom Enterprise. $999 of credit at $0.13 per CU gives roughly 7,684 CU per month. The concurrency cap is 256, RAM cap 256 GB, 500 datacenter IPs included with $0.60 per extra. The plan also includes Gold actor-store discounts, a named account manager, and monthly training hours.

The $0.13 CU rate is the cheapest published. For teams running heavy continuous workloads (multiple millions of pages a month, or large structured-extraction jobs), Business is the rational pick despite the headline number. The math is similar to Scale: at high compute volume, the unit-rate discount returns the higher subscription several times over.

How much does Apify cost for real workloads?

The honest answer is: it depends on what each page costs in Compute Units, and that varies tenfold depending on the actor. The math is easier with two reference points: lightweight (HTTP-only) and heavy (full browser rendering).

CU rate at $0.20 (Free / Starter). Heavy actor estimate assumes 4 GB RAM and full rendering with anti-bot routing.

Add the subscription floor and the proxy overage on top of that compute number. A team running 100,000 heavy-rendered pages a month on Starter is looking at roughly $100 of compute plus the $29 subscription plus any proxy overage, for a total comfortably north of $130. The same workload on Scale ($0.16 per CU) is $80 of compute plus $199 subscription, about $280, in exchange for higher concurrency and a quieter overage.

None of this is hidden. It is on the pricing page, in the docs, and in the actor configuration. It is just not summarized in a sentence, which is why the moment your prototype graduates into something with real traffic, the bill takes a step that nobody on the team modeled.

The crawlcrawl alternative: priced per page, not per Compute Unit

crawlcrawl is a crawler API built around a different pricing premise: one number per tier, one meter (pages), and every feature included. There is no separate Compute Unit rate, no proxy overage line, no actor RAM cap to watch, no rental marketplace. You buy a monthly page allowance, and the price on the table is the price on the invoice.

This section is the honest, detailed comparison, because a guide about Apify pricing should tell you what the realistic alternative actually does and where the trade-offs sit.

One meter, one number per month

A crawlcrawl tier sells you a flat page allowance. 1,500 pages on Free, 5,000 on Pro, 100,000 on Studio, 500,000 on Agency, 1,000,000 on Scale. JavaScript rendering, structured-data extraction, anti-bot routing, global routing across 190 or more countries, scheduled crawls, dataset storage, webhooks, the diff endpoint, the search API, and all eight content-intelligence actors are included at every paid tier with no per-feature surcharge.

The practical impact: there is no actor-RAM dial to misread, no Compute Unit multiplier to factor, no proxy overage to forecast. The bill is the tier. If your usage stays inside the page cap, your monthly invoice is identical month over month.

Pricing: every Firecrawl-class tier at half

crawlcrawl mirrors the Firecrawl volume ladder one for one and charges 50% at every paid tier. For Apify users comparing on the most common production workload (around 100,000 pages a month), the relevant line is Studio.

Apify estimates use the Compute Unit math above plus subscription floor. Light = ~1,000 pages per CU; heavy = ~200. Pricing verified 2026-05-23.

The number that matters most is the production line. A team running 100,000 heavy-rendered pages a month pays roughly $129 to $279 on Apify, depending on tier and actor weight, against $42 on crawlcrawl Studio. The annual gap on that workload alone is $1,000 to $2,800, before factoring proxy overage that often pushes the Apify number higher.

What you trade away by leaving Apify

An honest comparison names what is genuinely different, not only what is better. Apify's main strength is the actor marketplace: 31,000-plus pre-built scrapers covering specific sites and use cases. If your job is "scrape this exact Indeed listing format with anti-bot handling and a maintained schema", Apify's actor store probably has it. That is real value, and it is the obvious pick for a team whose workload is "many narrow vertical scrapers".

crawlcrawl is the general-purpose crawler API. You give it URLs and it returns clean markdown and structured signals. It is not an actor marketplace. The buyer who needs a turn-key Instagram scraper or a TikTok ranker should look at Apify. The buyer who needs a general crawler for documentation ingestion, RAG pipelines, content monitoring, or AEO audits is the one for whom crawlcrawl is built.

"We evaluated three crawlers before picking crawlcrawl. Structured-data extraction matters to us because we map customer-owned assets back to their security posture. Scheduled crawls plus webhooks gave us a live asset inventory with zero scripting. It paid for itself in the first month." — Rajesh Meta, Co-founder & CTO, Quick ZTNA

Predictable bills change how teams plan

The least obvious value of a single-meter price is what it does to engineering planning. With Compute Units, an engineer optimizing a scraper has to think about RAM, runtime, retry strategy, and proxy use, because each of those variables moves the bill. With per-page pricing, that engineer thinks about correctness and finishes faster. The math is done at the tier boundary, not on every actor.

This shows up in two places: budgeting (forecasts that match the invoice) and code review (no surprise costs from "small" actor changes). It is the kind of thing that does not appear on a pricing page but matters on the second sprint.

"We index documentation across forty vendor sites every week. crawlcrawl made it boring infrastructure, and that is the highest compliment I can give a tool." — Amit Tanwar, Founder, Networkers Home

Getting started: same first-curl as Apify

Signing up takes about the shape an Apify user expects. Create an account, get an API key prefixed crk-, and run your first scrape with one curl. No credit card on the free tier.

curl -X POST https://api.crawlcrawl.com/v1/scan \
  -H "Authorization: Bearer crk-YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"url": "https://example.com"}'

From there the production endpoints (async crawls, scheduled monitors, webhooks, structured-data extraction, the diff endpoint) are the same REST surface, documented in full on the API reference. The free tier of 1,500 pages a month is enough to run a representative evaluation against your actual workload.

Apify pricing FAQ

Is Apify free or paid?

Apify has a free tier that includes $5 of platform credit per month plus 5 datacenter proxy IPs. The credit is consumed at $0.20 per Compute Unit. Paid plans start at Starter ($29 per month) and go up to Business ($999 per month), with custom Enterprise above that.

What is a Compute Unit on Apify?

A Compute Unit equals 1 GB of actor memory running for 1 hour. A 4 GB actor running 15 minutes consumes 1 CU. The CU rate is $0.20 on Free and Starter, $0.16 on Scale, and $0.13 on Business. Your monthly platform credit is spent against this rate.

How much does Apify cost for 100,000 pages a month?

It depends on the actor's RAM and runtime per page. A lightweight HTTP actor at roughly 1,000 pages per CU puts 100,000 pages at 100 CU, or about $20 of compute on Starter. A heavier rendered actor at 200 pages per CU puts the same 100,000 pages at 500 CU or about $100. Add the subscription and proxy overage on top.

How do I use Apify for free?

Sign up for an account and use the $5 monthly platform credit. That gives roughly 25 Compute Units at $0.20 per CU. Treat it as evaluation, not a deployment plan; the credit goes fast on any recurring workload.

Why is Apify pricing so confusing?

Apify bills on five separate variables at once: the platform subscription, Compute Units, proxy IP overage, actor RAM caps, and concurrent run limits. The headline tier price is only the prepaid credit floor, not the final bill.

Can you make money with Apify?

Yes. Apify operates an Actor Store where developers publish scrapers others rent. Monetization is paid per usage of your actor and varies by your pricing model. This is separate from being an Apify customer; it is a development opportunity rather than a cost saving.

Is there a cheaper Apify alternative for general crawling?

Yes. crawlcrawl charges per page rather than per Compute Unit, with one tier price and all features included. Studio is $42 a month for 100,000 pages with 50 concurrent requests. For general-purpose crawling, RAG ingestion, or content monitoring, the price difference against an equivalent Apify workload is typically $80 to $250 a month on the same volume.

Which crawler should you choose?

If your workload is a specific scraper from the Apify Store that already exists, run with Apify; the actor marketplace is genuinely useful and saves you the build. If your workload is general-purpose crawling, documentation ingestion, RAG, content monitoring, or AEO audits, the per-page model on crawlcrawl is the simpler economics. If your monthly compute on Apify is heading past 1,000 CU, do the math against crawlcrawl Studio or Agency before renewing. If you need a custom volume or contract, talk to us.

curl -X POST https://api.crawlcrawl.com/v1/scan \
  -H "Authorization: Bearer crk-YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"url": "https://example.com"}'