Apify Pricing Explained (2026): What a Compute Unit Really Costs

Direct answer

Apify has four self-serve tiers, verified on its live pricing page in July 2026: Free ($0, $5 of prepaid usage a month), Starter ($29/mo), Scale ($199/mo), and Business ($999/mo), plus custom Enterprise. Your plan price is prepaid platform usage, metered in compute units (1 CU = 1 GB of RAM running for 1 hour) at $0.20/CU on Free and Starter, falling to $0.13/CU on Business. The real bill stacks three meters — compute units, proxy gigabytes ($8/GB residential), and any paid Actor's own fee — which is why costs feel unpredictable. Annual billing saves 10%.

Updated

Jul 12, 2026

Evidence

3 checks

Sources

3 source links

Target query

apify pricing

Evidence used

• Plan prices, compute-unit rates, proxy costs and concurrency limits verified against apify.com/pricing on Jul 12, 2026.
• Cost examples are modeled from Apify's own published unit rates, not quoted from vendor calculators; assumptions are stated in each table note.
• Corrects stale figures still circulating in competitor posts (e.g. $39 Starter and $0.30–0.40/CU), which no longer match the live page.

How we checked this

• Every price and unit rate here was read directly from the live Apify pricing page on Jul 12, 2026; Apify changes plan tiers and rates periodically, so recheck before committing budget.
• Compute-unit cost examples multiply Apify's published per-CU rate by an assumed RAM × runtime; real Actors vary with target-site complexity, concurrency and memory settings, so treat these as planning ranges, not quotes.
• This is a source-verified pricing explainer, not a ToolProven hands-on Apify benchmark — we have not yet run our own metered test suite on the platform, and we say so rather than implying first-party run data.

What your Apify plan fee actually buys

On Apify, your monthly plan fee is prepaid platform usage, not a flat all-you-can-scrape license. Starter's $29 buys $29 of consumption; when you exhaust it, overage bills to your next invoice. The tiers mainly buy three things beyond that credit: a lower compute-unit rate, higher RAM and concurrency ceilings, and better support.

This is the single most misunderstood thing about Apify pricing, and it is why the search results are full of confused cost estimates. You are not buying a fixed number of scrapes. You are pre-loading a usage balance and then drawing it down as your Actors run — much closer to cloud-compute billing than to a typical SaaS seat.

The practical consequence: two people on the same $29 Starter plan can have wildly different experiences. One runs a light daily scrape and never touches the ceiling; the other runs a memory-heavy browser automation across thousands of pages and blows through the prepaid balance in days. The plan name tells you almost nothing until you model your own workload.

What a compute unit is (and what one really costs)

A compute unit (CU) is Apify's core meter: 1 CU equals 1 GB of RAM running for 1 hour. An Actor using 2 GB of memory for 30 minutes burns 1 CU. At the Free/Starter rate of $0.20/CU, that run costs 20 cents; on Business it is 13 cents. The rate falls as you move up tiers, which is the main financial reason to upgrade beyond the higher ceilings.

The formula is worth memorizing because it is the whole game: cost ≈ (RAM in GB) × (runtime in hours) × (CU rate). A scraper you configure at 4 GB that runs for 15 minutes is 4 × 0.25 = 1 CU. The lever most people miss is memory: doubling an Actor's RAM allocation to make it faster also doubles its CU burn per hour, so “faster” is not automatically “cheaper.”

Watch for stale numbers when you compare guides. Several widely-ranked posts still quote Starter at $39 and CU rates of $0.30–0.40 — those were real once but do not match the live page in July 2026, where Starter is $29 and the Free/Starter CU rate is $0.20. When a pricing article and the vendor's own page disagree, trust the vendor page and note the date you checked.

The three meters that stack (this is why bills surprise people)

Apify's reputation for unpredictable pricing comes from meter stacking. A single job can draw on three separate meters at once: compute units for the run, proxy traffic for the requests (residential proxies are $8/GB — the priciest line), and, if you use a paid Actor from the Store, that Actor's own fee on top. Budget all three, not just the plan price.

Compute units are only the first meter. If your scrape needs residential proxies to avoid blocks, every gigabyte of proxy traffic is billed separately at $8/GB on Free and Starter (falling to $7/GB on Business). Datacenter proxy IPs are included in bundles (5 on Free up to 500 on Business) with extras priced per IP, and SERP proxy is billed per thousand queries. Proxy cost is frequently larger than compute cost on request-heavy jobs.

The third meter is the marketplace. Many of the best Store Actors are paid: some charge a flat monthly rental, others bill per result, per event, or per usage — and that fee is drawn against or added on top of your platform credit. A “$29 plan” running a pay-per-result lead scraper at scale is really paying $29 of platform usage plus the Actor's per-lead price plus proxy traffic. None of that is hidden, but you have to add it up yourself.

Worked examples: which plan fits which job

For most individuals evaluating Apify, the honest answer is: start Free to model your real workload, move to Starter ($29) once you exceed the $5 monthly credit, and only jump to Scale ($199) when the lower CU rate and higher concurrency actually pay for themselves. The plan is a function of compute + proxy volume, not of how many “scrapes” you think you need.

The Free tier's $5 of monthly prepaid usage is a genuine evaluation budget, not a toy: it is enough to run a prebuilt Actor across a few hundred pages and watch exactly how your CU and proxy meters move before you pay anything. Rented (paid) Actors are trial-limited on Free, so test with free Store Actors first.

The jump most people get wrong is Starter to Scale. Scale's value is not “more scrapes” — it is the CU rate dropping from $0.20 to $0.16 (a 20% cut on compute), concurrency rising from 32 to 128 parallel runs, and cheaper proxy tiers. If your monthly compute bill on Starter is small, those savings do not justify the $170 gap; if you are running large parallel jobs and burning proxy gigabytes, Scale can be cheaper in total than Starter-plus-overage.

What reviewers say about Apify's pricing

Apify scores well overall — 4.8/5 on Trustpilot (606 reviews) and Capterra (482 reviews), and 4.7/5 on G2 (checked July 2026) — but pricing is consistently its weakest point. G2's own top negative tags are “Pricing Issues” and “Expensive,” and the recurring complaint across platforms is not that Apify is overpriced so much as that the layered meters make bills hard to predict.

The pattern in the reviews maps exactly onto the meter-stacking above: people love the 50,000-plus prebuilt Actors and the managed infrastructure, then get surprised when a job's proxy or per-Actor fees dwarf the compute cost they budgeted for. Trustpilot's own summary of recent reviews flags “cost predictability for large workloads” as the standout concern.

Our read, as an affiliate that has not yet run its own metered bench on Apify: the ratings are strong and the platform is clearly capable, but treat the first month as a measurement exercise. Run on Free, watch all three meters, and size your plan from observed usage rather than the plan names. We will update this with first-party run costs once we bench it.

Related ToolProven pages