Apify Pricing: Free Credits, Actor Costs and Pay-as-You-Go Plans

Apify pricing explained: a platform budget, not unlimited scraping

Apify is a cloud platform for running ready-made web data collectors or custom automation programs called Actors. Its main paid plans combine a monthly subscription with a corresponding amount of prepaid platform usage. That means a Starter customer pays $19 per month and receives $19 to spend on eligible Actor and platform activity, while a Scale subscriber pays $199 and receives a $199 usage allocation. A subscriber who consumes more than the included amount can allow additional pay-as-you-go charges.

The official November 2026 pricing grid shows Free at $0 with $5 monthly use, Starter at $19, Scale at $199, Business at $999 and custom arrangements by quotation. Those are subscription prices, not fixed prices for a certain number of website pages. Different actors may price outcomes, runtime or both. A scraper that uses browser rendering, frequent retries and residential proxies can cost much more per usable record than a simple HTML request. Price estimates therefore need a workload, not only a plan name.

Free: $5 monthly usage without a credit card

The Free plan allows a small proof of concept on Apify Store or a custom Actor with a monthly $5 allowance. The platform lets users select a ready-made collector, provide an input such as a permitted list of URLs, run it and download or access the result data. Five dollars can be enough to learn how the Console, run logs and datasets fit together, but it does not promise a particular volume of valid rows. Actor choice and resource use determine how far the allowance stretches.

Apify’s subscription documentation explains that Free stops further usage after its prepaid allocation is consumed and resumes when the next cycle begins. Ordinary pay-as-you-go overage is a paid-plan option, not a hidden free-plan debt. The public pricing comparison lists a maximum of five concurrent Actor runs and $0.20 per compute unit. Anyone using the free tier to validate a data product should keep jobs short and record the actual usage cost, output size and error rate before drawing conclusions about production budgets.

Starter: $19 per month with $19 to spend

Starter is the lowest paid tier at $19 monthly. Its subscription includes $19 of prepaid usage in that billing period, and its standard compute price remains $0.20 per CU. The plan’s listed simultaneous run limit rises to 32, and Apify advertises chat support and a Bronze discount on eligible Store activity. The essential difference from Free is room to run more tasks and the option to pay for usage over the prepaid amount when the account limit is raised.

A common mistake is describing the plan as $19 of platform credit on top of a separate $19 monthly fee that is never used toward workloads. The actual model is a subscription containing that prepaid allowance. With the default spending limit, a running workload may stop once it reaches the current cap rather than automatically accumulate overage. The vendor’s billing help has an older illustrative Starter amount that differs from the live pricing page; use the current $19 plan page and the account’s Billing screen for the actual invoice.

Scale: $199 monthly and lower compute pricing

Scale lists $199 per month with $199 of prepaid usage, a compute rate of $0.16/CU and 128 simultaneous Actor runs in the current grid. It adds higher aggregate resource limits, Silver Store discounting and priority chat support. For a team running scheduled jobs across many permitted sites, that larger allocation can simplify monthly budgeting and lower the nominal cost of compute-heavy operations. Scale does not transform every Actor into a fixed-price service.

Moving to Scale makes sense when regular consumption and concurrency justify the larger commitment. If a team uses $40 of resources one month and $80 the next, a $199 subscription may waste credit because unused prepaid usage expires. If it consistently consumes several hundred dollars with peak parallel jobs, the lower CU rate and expanded capacity become relevant. Compare real invoices for at least a few representative billing periods before relying on theoretical compute-unit savings.

Business: $999 per month for high-volume workloads

Business lists $999 each month with matching prepaid usage, a $0.13 per compute unit rate and up to 256 concurrent Actor runs. It also advertises larger memory and aggregate runtime resources, Gold Store discounts and an account manager. The tier is aimed at organizations that operate large collections, dedicated data pipelines or many simultaneous jobs. It should not be evaluated merely by counting individual web pages processed in one test run.

A high-volume company should examine its workload shape. Two teams could both consume $999 of monthly service, but one may pay primarily for low-compute per-event Actor results while another pays for long browser sessions and proxy bandwidth. Reduced CU pricing benefits the second workload more directly. Enterprise contracts are separate: companies requiring bespoke SLAs, managed development or special delivery should ask for written service and pricing terms instead of treating Business as an implicit guaranteed-output service.

Annual billing and the advertised ten-percent discount

Apify’s pricing selector advertises a ten-percent saving for annual billing. Applying that discount to the displayed subscription sticker rates gives mathematical monthly equivalents of $17.10 for Starter, $179.10 for Scale and $899.10 for Business, before taxes and any other charge. Those are arithmetic equivalents, not monthly cancel-at-will invoices. The prepaid allowance and annual billing arrangement should be verified at checkout because subscription discounts do not necessarily mean usage consumption itself receives a matching reduction.

The vendor documents mid-cycle upgrades and adjustments to prepaid usage based on how many days remain. It also warns that switching away from an annual commitment does not automatically refund the remainder. If the workload is uncertain, a month or two of measured use on Starter or Scale can be more informative than immediately locking in annual billing. Buyers should record renewal timing, invoice currency, VAT, cancellation obligations and which discounts affect Actor usage as distinct line items.

Compute units: one gigabyte of RAM for one hour

Apify defines a compute unit, or CU, as approximately one gigabyte of RAM used for one hour of execution. For a workload allocated 2 GB of memory for 30 minutes, compute consumption is roughly one CU. If that same job runs for two hours, it uses about four CU. This is a resource-billing illustration, not a promise that the program will obtain any particular amount of information while running. Performance depends on target pages, browser work and scheduling.

At $0.20/CU on Starter, 100 CU of metered compute would cost $20 before any other usage. At $0.16/CU on Scale it would be $16, and at $0.13/CU on Business it would be $13. The same compute count can have dramatically different total cost when the Actor also charges event fees, proxy traffic or storage. Do not price an Actor from its memory setting alone; inspect the store page’s Pricing section and the detailed cost breakdown of an actual run.

Pay-per-event Actors can change the whole calculation

Apify Store includes programs created by Apify and independent developers. Some are billed for a successful result, for starting a run or for another event that the developer defines. Others charge only for platform resources, and certain workloads combine event prices with extra platform usage. The official documentation says many pay-per-event Actors bundle infrastructure into the event charge, but not all do. This makes a general platform-wide cost-per-record figure misleading.

For a concrete comparison, imagine an Actor charging $2.30 for 1,000 results and another program using four compute units plus proxy traffic to return the same number of records. The first has a visible event rate but may require eligibility conditions; the second’s cost depends on resources. The Apify pricing page itself uses actor-specific illustrations rather than claiming one universal record price. Review the per-event definitions, maximum charges, output conditions and possible charges for failed attempts before scheduling.

Proxy, storage and data transfer charges

The official pricing table lists residential proxy traffic at $8/GB on Free and Starter, $7.50/GB on Scale and $7/GB on Business. It also lists datacenter IP allowances and rates for Google search-result proxies, the Unblocker service, storage operations and network transfer. Not every Actor requires each component. A simple extraction from a site that permits ordinary requests may use fewer resources than a workflow executing a real browser repeatedly and transferring many image-heavy pages.

Proxy traffic should be measured separately from downloaded result rows. A scraper may transfer several megabytes of page resources to obtain a few kilobytes of useful data. Storage charges may appear after execution when large datasets remain accessible, and transfer fees can matter for repeated exports. Use Apify’s Billing → Pricing screen to examine exact current rates under the account’s tier and inspect Historical usage for the resources used by each Actor.

Concurrency, memory and automation limits

The live grid lists maximum concurrent Actor runs of five on Free, 32 on Starter, 128 on Scale and 256 on Business. It also gives different RAM allocations and total running-memory limits. A job that only needs one nightly crawl does not gain much from a 256-run allowance, but a company refreshing many independent sources during the same window could benefit. Parallelism may reduce elapsed time while increasing simultaneous resource use.

Apify also documents maximum numbers of schedules, tasks and webhooks. Those platform limits are a separate question from monthly usage credits. A poorly configured schedule can trigger several costly runs that fetch duplicate data; a narrow concurrency allowance can instead queue work and delay delivery. Before selecting a tier, record how often each dataset must refresh and how many Actors must run at the same time. Set per-Actor maximum charge as well as overall monthly limits where supported.

Spending caps, overage and expired credit

Apify defaults the platform spending limit to the prepaid amount for paid plans according to its current billing guide. The Limits tab lets a user raise the cap, at which point additional consumption can appear as a pay-as-you-go charge. Free has a more restrictive stop at the included allowance. This is useful protection, but it is not a replacement for looking at each Actor’s own event pricing and execution logs.

Prepaid usage does not roll over. If a team pays for Scale and consumes only $80 of its $199 allocation, it should not expect the unspent difference to be available next month. Apify’s historical usage view may also show charges before discounts, so a team should reconcile raw consumption with the final invoice. A sensible budget includes a monthly cap, an alert threshold, a run-level limit and a process for pausing the schedule when output quality drops or a data source changes.

How to estimate cost per usable record

A scraped page is not the same as a useful row. For a market-research project, the useful result might require a correct business name, permitted contact field, current timestamp and a valid source URL. Measure the number of complete, nonduplicate records that pass these checks. Then divide the total charged usage by that accepted count. A nominal $0.002 per generated row can be much more expensive if half the fields are blank or records cannot legally be reused.

Start with a small, authorized collection, note the Actor version and configuration, then inspect runtime, CU use, event charges, proxy bandwidth, output count and data-quality rejection rate. Repeat at a larger permitted sample only if the results remain stable. Include maintenance time in the calculation: when a target website changes its structure, an Actor may begin producing empty data while still consuming resources. That operational cost is real even if the platform subscription stays the same.

Examples for SEO, ecommerce and AI data projects

An SEO team might collect publicly available metadata from approved websites for a one-time content audit. A small predefined list and simple HTML collection could fit the free allowance or Starter. An ecommerce research team refreshing thousands of permitted product pages daily may need schedules, more concurrency, costlier proxy traffic and systematic duplicate handling. Its recurring usage can quickly change the value of Scale or Business.

An AI development team might run a crawler to turn permitted site content into a structured dataset for search or retrieval. Costs then include execution, storage and the downstream model or vector database, which is not part of Apify pricing. If only clean Markdown pages are required, a focused service such as Firecrawl may be simpler. If the workload includes browsing actions, multiple marketplaces and custom logic, Apify’s Actor architecture may justify the operational overhead.

Plan choice: choose by workload and maintenance capacity

Starter fits a developer or analyst who wants to validate a small workflow and access paid-account overage controls. Scale fits teams that consume meaningful monthly resources and need many concurrent scheduled runs. Business fits sustained high-volume operations that can use lower compute rates, more capacity and an account manager. Custom plans fit specific contractual delivery, support or engineering needs. The Free plan is most useful for proving the mechanics with no payment card.

The economic break-even is not simply a CU price ratio. A program priced per event may benefit from Store discounts rather than compute rates. Proxy and other resources have their own tier schedules. Compare the total cost of the same approved jobs on current prices, including annual commitment, unused prepaid credit and variable work. A plan that looks larger can be cheaper at scale, but a plan with unused fixed credit can be expensive for an irregular research project.

What to check before you upgrade or cancel

Before upgrading, check that the Actor’s costs are recurring and predictable, rather than a brief unusual spike. Apify’s help explains that users can raise a usage limit temporarily without moving to the next plan, which may be preferable for one expensive run. If you do upgrade mid-cycle, charges and prepaid usage are prorated based on the change date while the billing cycle itself stays aligned to the earlier start.

Before downgrading or canceling, turn off schedules, inspect active runs and export datasets you still have permission to retain. Confirm billing currency, expiration of unused credit, retention and the status of any annual commitment. A platform plan should be chosen together with a data governance plan: know which sites authorize the collection, what personal data may appear and how long the organization is allowed to keep it.

Pricing research and verification status

This pricing article uses Apify’s public USD price grid, product documentation for subscriptions, Store Actors, platform limits and billing help, checked November 5, 2026. I have not created an Apify paid account, purchased an Actor, verified live invoice calculations, tested an extraction workload or asked sales for a custom quote. Apify’s support pages occasionally retain older illustrative amounts, so the live plan grid and the account’s confirmed checkout are the appropriate references for a final purchase.