Apify pricing

About

Apify is a web scraping and browser-automation cloud built around Actors — serverless programs that run in Docker containers and can be scheduled, chained, and called via API. The platform pairs a developer runtime (Crawlee, the open-source scraping library, plus proxies and storage) with Apify Store, a marketplace of 35,000+ ready-made scrapers that creators publish and monetize. Customers either build their own Actors or rent/run Store Actors, and pay for the underlying platform resources those runs consume.

Apify serves a broad spectrum: solo developers and indie scraper authors at one end, and enterprises sourcing data for generative AI, AI agents, lead generation, and market research at the other. The company positions itself as full-stack infrastructure for “data for AI” — it advertises that Store creators were paid over $1M last month, with many earning more than $3k.

Competitively, Apify sits among web-scraping and data-extraction platforms (Bright Data, ScrapingBee, Zyte) but differentiates with its two-sided Actor marketplace and a usage metric — the compute unit — that abstracts memory and CPU into a single billable number.

Pricing summary : prepaid platform usage on compute units, plus marketplace Actor fees

Apify uses a hybrid model: a flat monthly subscription that includes a matching dollar amount of prepaid platform usage, with any overage billed pay-as-you-go to the next invoice. Platform usage is metered across four components — compute units, data transfer, proxy, and storage operations — and the headline lever is the compute unit (CU), defined as 1 GB of RAM allocated for 1 hour. The per-CU rate falls as you move up tiers, from $0.2/CU (Free/Starter) to $0.16 (Scale) and $0.13 (Business).

The billing dimensions are:

• Subscription tier — Free ($0, $5 prepaid), Starter ($29), Scale ($199), Business ($999), Enterprise (custom). Each tier’s prepaid usage equals its monthly fee (except Free’s $5 grant), and each raises Actor RAM and max-concurrent-run ceilings.
• Compute units (CU) — 1 GB RAM × 1 hour = 1 CU, priced $0.2 → $0.13/CU by tier; the dominant cost of most Actor runs.
• Proxy — residential ($8 → $7/GB), datacenter (included IPs then $1 → $0.6/IP), and SERP ($2.5 → $1.7 per 1,000) usage, all drawn from prepaid usage.
• Storage operations — per-1,000 reads/writes/lists and timed GB-hour storage on Dataset, Key-value store, and Request queue.
• Data transfer — external ($0.20 → $0.18/GB) and internal ($0.05 → $0.04/GB).
• Marketplace Actor fees — Store Actors layer their own charge on top of platform usage: pay-per-event (e.g. per result), pay-per-usage (platform cost only), or rental (flat monthly, being sunset in 2026).

What makes this different: Apify meters a single synthetic unit (the CU) rather than seats or raw API calls, and runs a creator economy on top where Actor authors keep 80% of pay-per-event revenue net of platform costs.

Pricing by product

Apify Platform — subscription plans (monthly, USD)

Annual billing is 10% cheaper than the monthly prices above. Students get 30% off Starter/Scale; startups get 30% off Scale; nonprofits get a personalized discount.

Creator Plan — for Actor developers

Residential proxy capped at 10 GB/month and SERPs at 10,000/month on the Creator Plan; $500 bonus granted once per account.

Platform usage rates (drawn from prepaid usage, then overage)

Marketplace Actor pricing models (buyer-facing)

Add-ons (purchased on top of a plan): concurrent runs $5/run, Actor RAM $1/GB, datacenter proxy from $0.6/IP, priority support $100, personal training $150/hour.

Sales motions across products: PLG / self-serve for Free, Starter, Scale, Creator, and Store Actors; sales-led for Business overage at scale and Enterprise custom plans.

Hidden costs : where the compute-unit and proxy meters quietly add up

The sticker price of an Apify plan is only the prepaid floor — the real bill is what your Actor runs consume in compute, proxy, storage, and transfer. Because the headline meter is the compute unit (memory × runtime), the two biggest surprises are heavy browser-based scrapers that allocate large RAM, and residential-proxy GB on sites that block datacenter IPs.

Archetype 1 — a solo developer on Starter running a daily browser scraper. A Playwright-based Actor at 4 GB RAM taking ~15 minutes per run consumes ~1 CU per run; run hourly across the month that is ~720 CU.

The $29 prepaid covers only the first ~145 CU; everything past that is pay-as-you-go overage on the next invoice. At this volume the developer is paying ~6× the plan price in usage — the classic signal to move up to Scale, where the CU rate falls to $0.16.

Archetype 2 — a team on Scale scraping a heavily protected e-commerce site. Proxy, not compute, dominates here.

Here the $450 residential-proxy line alone exceeds the plan price by more than 2× — proof that on protected targets Apify’s bill is a proxy-cost story, not a compute story. Teams that under-budget proxy GB get the worst bill-shock surprises, and the fix is usually swapping to datacenter IPs where the target allows and capping spend in Billing.

Want to estimate your own Apify bill? Use the Apify pricing calculator to model your monthly cost based on compute units, proxy GB, storage operations, and data transfer.

Pricing evolution : from a credits model to cheaper compute units and a rental sunset

Apify’s pricing has moved in an unusual direction for a usage-based platform: down. Plan prices and per-unit rates have fallen repeatedly since 2023, and the company is now simplifying its marketplace by retiring an entire pricing model.

Cadence

Tracked range: 2022-02–2026-06. Quarters not listed above were verified stable (0 price changes, 0 SKU additions). Apr–Sep 2025 Wayback snapshots rendered as JS skeletons, so the September CU cut is dated from the PricingSaaS report and the current live page.

Notable changes

• 2023-06 — Plans renamed to Free/Starter/Scale/Business and the compute unit (1 GB RAM/hour) formalized as the primary meter at $0.4/$0.4/$0.3/$0.25 per CU (apify.com/pricing, Wayback 2023-06).
• 2023-09 — Creator Plan introduced for Actor developers: $1/mo with a one-time $500 usage bonus (apify.com/pricing/creator-plan, Wayback 2023-09 / 2024-05).
• 2025-01 — Scale plan observed cut from $499 to $199/month, with residential-proxy and SERP rates lowered; CU rates unchanged (apify.com/pricing, Wayback 2025-01 vs 2024-05).
• 2025-09-13 — Compute-unit rates cut roughly 20–25% across all tiers to today’s $0.2/$0.2/$0.16/$0.13, and Starter moved to $29 (PricingSaaS change record; corroborated by the live pricing page).
• 2026 — Rental Actor pricing model scheduled for sunset: no new rentals from April 1, full retirement and migration to pay-per-usage on October 1 (docs.apify.com, Actors-in-Store).

The rental sunset in detail

Rental was one of four Actor pricing models on Apify Store: a creator could charge a flat recurring monthly fee (after a free trial) on top of normal platform usage, with the fee deducted automatically from a buyer’s prepaid usage. It was the closest thing Apify had to a pure subscription inside an otherwise usage-metered marketplace — and it is being removed. Per the Actors-in-Store docs, April 1, 2026 blocks publishing new rental Actors or repricing existing ones, and October 1, 2026 retires rental entirely, migrating remaining Actors to pay-per-usage. The practical effect is a cleaner two-axis story for buyers — platform usage plus an optional pay-per-event charge — and the removal of an opaque flat fee that sat awkwardly next to the consumption meters. It is a roadmap and trust event worth watching: creators who built recurring revenue on rental must repackage as pay-per-event before the deadline.

What’s unique : the compute-unit meter and a two-sided Actor marketplace

The compute unit abstracts memory and time into one number. Instead of charging per page, per row, or per API call, Apify bills a synthetic unit — 1 GB of RAM for 1 hour — so the same meter covers a lightweight Cheerio crawler and a heavy Playwright browser job. This is a hybrid usage model where the value metric is infrastructure consumption, which is honest (you pay for what you run) but hard to forecast (you must test-run to estimate).

A creator economy sits on top of the platform. Apify Store is two-sided: developers publish Actors and monetize them via pay-per-event, pay-per-usage, or (soon-deprecated) rental, while Apify takes a platform cut and pays creators 80% of pay-per-event revenue. Few infrastructure vendors run a marketplace where third-party pricing is layered on their own usage meter.

Prepaid usage equals the plan fee, and it expires. Every paid tier’s prepaid balance is exactly its monthly price — $29 of Starter buys $29 of usage — which makes the plan fee feel like a usage wallet rather than a seat license. The catch: unused prepaid usage expires at cycle end and never rolls over.

Pay-per-event runs can be hard-capped. For pay-per-event Actors, buyers set ACTOR_MAX_TOTAL_CHARGE_USD; the Actor terminates gracefully at the limit and never charges produced events beyond it — a rare per-run cost ceiling in usage-based billing.

Prices have fallen, not risen. Unusually for a metered platform, Apify has cut Scale ($499→$199), Starter ($49→$29), and CU rates (~20–25% in 2025) — the opposite of the price-creep most usage vendors drift toward.

Strengths & weaknesses

Billing UX : how Apify exposes usage, limits, and overage controls

Apify exposes usage and spend through several named controls:

• Prepaid usage meter — each cycle starts with a dollar grant equal to the plan fee; the Billing section of Apify Console shows current-period and historical usage drawn down against it.
• Pay-as-you-go overage — paid plans keep running past the prepaid balance and bill the excess to the next invoice; the Free plan is hard-blocked until the next cycle instead.
• Usage limit / billing cap — paying users configure a maximum spend in Billing; the platform stops charging overage once the configured limit is reached.
• Per-run Usage column and breakdown — the Runs section shows each run’s CU and dollar usage, with a detailed four-component breakdown (compute, data transfer, proxy, storage) behind the ”?” icon on the run detail page.
• ACTOR_MAX_TOTAL_CHARGE_USD — for pay-per-event Actors, users set a maximum cost per run; the Actor terminates gracefully at the limit and never charges produced events beyond it.
• Historical usage tab — PPE charges and rental deductions appear in the Historical usage view and the per-Actor breakdown within the Current period tab.

Strategic wins : pricing decisions that compounded Apify’s position

1. Picking a single synthetic meter (the compute unit)

By billing one unit — memory × runtime — Apify made every Actor, from a one-line HTTP fetch to a full browser session, priced on the same axis. That consistency lets the company publish one rate card and lets buyers reason about cost without learning a per-Actor pricing language. It is the same simplification logic behind choosing one durable value metric instead of many.

2. Making the plan fee a prepaid usage wallet

Setting each tier’s prepaid balance equal to its monthly fee removes the “am I paying twice?” objection — the subscription is the usage budget. This is a clean expression of a prepaid-credit model and nudges buyers to pick the plan whose wallet matches their expected spend, then upgrade for the lower per-CU rate.

3. Building a two-sided creator marketplace on top of the meter

The 80% pay-per-event revenue share turned Apify Store into a flywheel: more creators publish Actors, which attract more buyers, whose usage funds more creator payouts (over $1M in a single month). The marketplace is a moat competitors selling raw scraping infrastructure don’t have, and it converts the platform’s usage-based billing rails into a distribution channel.

4. Cutting prices to chase the AI-data wave

Rather than raise rates, Apify lowered Scale, Starter, and CU prices as it repositioned around “data for AI agents.” Cutting the meter while the addressable market expanded is a volume bet — cheaper units drive more runs, which is the right play when usage is elastic and the buyer is a developer counting CUs.

Areas to improve : where Apify’s pricing still creates friction

1. Make CU cost estimable before the first run

The biggest complaint is that you cannot predict an Actor’s CU consumption without running it. Apify could publish a typical-CU-per-1,000-results band on each Actor’s page (derived from historical runs), turning the test-run ritual into an up-front estimate and reducing cost unpredictability at purchase time.

2. Soften the prepaid-expiry cliff

Usage that expires at cycle end punishes spiky workloads — a scraper that runs hard one week and idle the next forfeits prepaid balance. A modest rollover (e.g. carry up to one cycle of unused usage) would reward commitment without undermining the wallet model, similar to how credit-grant designs handle burst usage.

3. Give residential proxy its own budget control

Because proxy GB is the top bill-shock source, a dedicated per-dimension cap (separate from the global spend limit) would let teams bound proxy spend specifically. Surfacing a proxy-GB forecast in Billing would also help finance teams attribute cost to the right scraping job.

4. Publish indicative Enterprise economics

Enterprise is entirely custom — no published CU rate, concurrency, or floor. A “starts at” CU rate or a typical contract band would help midmarket buyers self-qualify before booking a sales call, the same transparency gap that slows many sales-led usage deals.

Key takeaways

1. A synthetic meter can unify a sprawling catalog. Apify prices 35,000+ Actors on one compute-unit axis instead of bespoke per-Actor units — proof that the right value metric is the one that generalizes across your whole product surface.
2. Make the subscription double as the usage budget. Setting prepaid usage equal to the plan fee kills the “paying twice” objection and turns tier selection into wallet-sizing. It is a clean prepaid-credit pattern other usage vendors can copy.
3. Cutting your meter can be a growth lever. Apify lowered Scale, Starter, and CU rates rather than raising them, betting cheaper units drive more runs — a volume strategy that works when buyers are price-sensitive developers.
4. Hard per-run caps build trust in metered billing. ACTOR_MAX_TOTAL_CHARGE_USD lets buyers bound a single run’s cost, removing the fear that a runaway job produces a runaway invoice.
5. Retire pricing models that confuse buyers. Sunsetting rental in favor of pay-per-usage simplifies the Store’s pricing story — a reminder that fewer, clearer models often beat more options.

UBP implications

1. Memory × time is a viable universal meter for compute-heavy products. Apify shows that a single GB-hour-style unit can absorb wildly different workloads, which matters as more AI and agent products bill on consumed infrastructure rather than seats.
2. Prepaid-equals-plan-fee blurs the line between subscription and usage. This hybrid structure is a template for vendors who want recurring revenue and metered upside without two separate line items — but the expiry rule is the lever that decides how customer-friendly it feels.
3. Marketplaces can layer third-party pricing on a host meter. Apify’s 80% revenue share over pay-per-event events shows usage-based rails can become a monetization platform for an ecosystem, not just a billing method for one vendor.

Sources

• Apify pricing page — plan tiers, compute-unit rates, proxy/storage/transfer rates, add-ons (accessed 2026-06-03)
• Apify Creator Plan — $1/mo Creator Plan and $500 usage bonus (accessed 2026-06-03)
• Compute units & platform usage (docs) — CU definition (1 GB RAM/hour), memory/CPU allocation, four usage components (accessed 2026-06-03)
• Actor pricing models (docs) — pay-per-event, pay-per-usage, rental, and the rental sunset schedule (accessed 2026-06-03)
• Pay-per-event monetization (docs) — per-result event, ACTOR_MAX_TOTAL_CHARGE_USD, 80% creator revenue share (accessed 2026-06-03)
• Apify changelog — product and pricing announcements (accessed 2026-06-03)
• Apify pricing — Wayback Machine snapshots — historical plan and rate verification, 2022–2025 (accessed 2026-06-03)
• Compare with peer web-scraping platforms in the blueprint corpus.

Bottom line

Apify turns web scraping into a single, honest meter — the compute unit — wrapped in a prepaid-usage wallet and a two-sided Actor marketplace, and it has spent the last three years making that meter cheaper rather than dearer. The trade-off is forecastability: you only learn an Actor’s true cost by running it, and residential-proxy GB can quietly outrun the plan fee on protected targets. With the rental model retiring in 2026, Apify’s pricing is converging on two clean axes — platform usage plus optional pay-per-event — that make it one of the more legible usage-based stories in the blueprint corpus.

Pricing timeline : Major events on a vertical axis

Each milestone below corresponds to a public pricing change, product launch, or material adjustment. Major events use a filled marker; minor adjustments use a faded one.

Current snapshot — prepaid usage on compute units

Jun 2026

Plans: Free $0 ($5 prepaid), Starter $29, Scale $199, Business $999, Enterprise custom; Creator $1/mo with a $500 6-month bonus. Compute-unit rate steps from $0.2 to $0.13/CU by tier. Annual billing is 10% cheaper. Store Actors layer pay-per-event, pay-per-usage, or (sunsetting) rental fees on top, with creators keeping 80% of pay-per-event revenue.

Rental Actor pricing model sunset announced

Jan 2026

Apify announced it is retiring the rental Actor pricing model: from April 1, 2026 no new rental Actors can be published or repriced; on October 1, 2026 all remaining rental Actors are fully retired and migrated to pay-per-usage. Buyers and creators are pointed to the #project-rentals Discord channel.

Compute-unit rates cut ~20–25%; Starter cut to $29

Sep 2025

Apify reduced compute-unit rates across all tiers (reported ~20–25% by PricingSaaS, 2025-09-13) to the current $0.2/$0.2/$0.16/$0.13 per CU, and the Starter plan moved from $49 to $29/month. Wayback snapshots for Apr–Sep 2025 were JS skeletons, so the exact date relies on the PricingSaaS report plus the live page.

Scale plan cut from $499 to $199; proxy rates lowered

Jan 2025

Between mid-2024 and January 2025 the Scale plan dropped from $499 to $199/month (Starter held at $49, Business at $999). Residential proxy fell to $8/$8/$7.5/$7 per GB and SERPs to $2.5/$2.5/$2/$1.9 per 1,000. Compute-unit rates were unchanged at $0.4/$0.4/$0.3/$0.25.

Creator Plan launched for Actor developers

Sep 2023

Apify introduced the Creator Plan: $1/month (total $6 for the 6-month plan) with a one-time $500 platform-usage bonus usable in the first six months, aimed at developers building and publishing their own Actors, with limited Store access (universal Actors only).

Rename to Starter/Scale/Business and the compute-unit meter

Jun 2023

Plans were renamed Free / Starter ($49) / Scale ($499) / Business ($999) / Enterprise, and the compute unit (1 GB RAM/hour) became the headline meter at $0.4 / $0.4 / $0.3 / $0.25 per CU by tier. Residential proxy was $13/$13/$11/$10 per GB and SERPs $3 per 1,000.

Credits model with Personal and Team plans

Feb 2022

Early Apify pricing offered Free, Personal ($49/mo), Team ($499/mo, marked Recommended), and Enterprise. Plans bundled platform usage credits ($5 / $49 / $499) used for Actors, storage, and proxy; residential proxy was on-request and the free plan gave 4 GB Actor RAM.