Apify vs Bright Data vs Oxylabs: Managed Scraping Platforms Compared (2026)

apify is the developer-friendly platform with a marketplace of 3000+ pre-built scrapers (called actors). bright data has the deepest catalog of pre-scraped datasets and the most complete proxy + unlocker infrastructure. oxylabs is the enterprise-grade api stack with the cleanest docs and the strongest sla. they overlap but serve different buyers. this is the may 2026 comparison with prices from each dashboard.

the matrix

prices change. snapshot is may 1, 2026.

apify: the developer platform

apify is the only one of the three that’s a real platform-as-a-service for scraping. you write a docker container that scrapes some target, push it to apify, and they run it on their infrastructure. they call this an “actor.”

3000+ public actors exist already. amazon scraper, instagram scraper, google maps scraper, linkedin scraper, twitter scraper, facebook ads library scraper, youtube scraper. most are maintained by apify themselves or trusted community devs.

from apify_client import ApifyClient

client = ApifyClient("YOUR_API_TOKEN")

# run the official google search scraper actor
run = client.actor("apify/google-search-scraper").call(run_input={
    "queries": "best mobile proxies singapore",
    "maxPagesPerQuery": 3,
    "resultsPerPage": 100,
    "countryCode": "us",
})

# fetch results
for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(item["title"], item["url"])

5 lines of code, no scraper to maintain. you pay the platform usage (compute units + bandwidth) plus the actor’s per-result fee where applicable.

apify pricing is denominated in compute units (cu). 1 cu is roughly 1gb-hour of memory + cpu. a typical google search scrape uses 0.1-0.5 cu. the $49/month starter plan gives you 49 cu, which goes far for hobbyist scraping.

what’s bad: apify’s residential proxy pool is smaller than bright data’s or oxylabs’. for protected sites, you’re better off pairing apify’s compute platform with an external residential proxy. they support this directly in actor inputs.

we cover apify in our web scraping apis for developers 2026 article.

bright data: the data and infrastructure giant

bright data is the largest of the three by revenue and pool size. their proxy network is the deepest, their web unlocker handles the meanest sites, and their dataset catalog is unmatched.

three product lines:

(1) proxy network. residential, datacenter, isp, mobile. all the standard tools.

(2) managed apis. web unlocker, serp api, scraping browser, data collector. these are bright data’s “we’ll do the scraping for you” stack.

(3) datasets. pre-scraped data sold by record. you don’t run a scraper, you just pay for the rows. linkedin profiles, amazon product data, zillow listings, instagram public posts.

import requests

# trigger a managed scrape via web unlocker
resp = requests.get(
    "https://api.brightdata.com/dca/trigger",
    headers={"Authorization": "Bearer YOUR_TOKEN"},
    json={
        "url": "https://www.target-site.com/products/123",
        "render": True,
    },
)

bright data’s strength is breadth + scale. if you’re scraping at million-page-per-day volumes or need 100% success on sites with serious anti-bot, bright data is the lowest-risk pick.

starter plan is $499/month with usage-based billing on top. expensive entry. for under $1000/month spend, you’ll feel the price more than the technical advantage.

we go deep on bright data in our proxy providers ultimate guide.

oxylabs: the enterprise scraping api stack

oxylabs is the cleanest of the three for engineers who want straightforward apis with no marketplace, no datasets, and no actor system. their core scraping products:

(1) web scraper api. send a url, get back html or auto-extracted json.

(2) serp scraper api. send a query, get back parsed search results.

(3) ecommerce scraper api. send a url to amazon/ebay/walmart/etc, get back parsed product json.

(4) real estate scraper api. send a url to zillow/realtor/redfin/etc, get back parsed listing json.

import requests

resp = requests.post(
    "https://realtime.oxylabs.io/v1/queries",
    auth=("USER", "PASS"),
    json={
        "source": "ecommerce_product",
        "url": "https://www.amazon.com/dp/B0CHX1W1XY",
        "parse": True,
    },
)
print(resp.json()["results"][0]["content"])

response includes parsed product fields ready for ingestion. pricing is per request, $0.001-$0.005/req depending on source and volume tier.

oxylabs’ strength is api cleanliness and reliability. their dashboards are fast, their docs are excellent, and their support team is 24/7 with strong sla. for an enterprise data team that wants “we send urls, you give us clean data,” oxylabs is the easiest sell internally.

what’s bad: no marketplace. if you want a tiktok scraper or a youtube scraper, oxylabs doesn’t have one. you’d need to build it yourself using their proxy network.

use case decision tree

(1) you want to scrape a major site (google, amazon, instagram, linkedin) and don’t want to write or maintain a scraper. all three work, but apify’s actor catalog is the deepest. if the actor exists, apify wins on developer time saved.

(2) you want to buy pre-scraped data without running scrapers at all. bright data datasets is the broadest catalog. apify also has dataset listings on their store but volume is smaller.

(3) you scrape protected sites at high volume and need near-perfect success. bright data web unlocker. expensive but the success rate justifies the spend.

(4) you want clean apis with predictable pricing for major ecommerce or real estate sites. oxylabs is the cleanest.

(5) you’re building your own scraper and want a hosting platform that handles concurrency, retries, and storage. apify is the only one of the three. bright data and oxylabs sell apis, not platforms.

(6) you have an enterprise procurement process and need vendor-grade contracts, sla, and account managers. oxylabs and bright data both qualify. apify can but is more startup-coded.

actual cost comparison

pricing varies by use case. some realistic monthly bills based on real customers we’ve talked to:

bright data wins on extreme-volume protected scraping (the unlocker is genuinely cheaper at scale). apify wins on small-to-medium custom scraping where actor reuse pays off. oxylabs wins on clean enterprise spend predictability.

developer experience

apify: best of the three. apify cli, sdk in python and node, actor sdk for building your own, dashboard with logs, dataset viewer, scheduler, alerts. you can ship a custom scraper in an afternoon.

bright data: most surface area, steepest learning curve. multiple products with overlapping features. dashboard is feature-rich but takes time to learn. docs are thorough but scattered.

oxylabs: cleanest of the three for the api-only flow. one endpoint pattern (POST /v1/queries with a source parameter) covers most use cases. minimal cognitive load.

marketplace quality on apify

since apify’s marketplace is the unique selling point, here’s a snapshot of what’s actively maintained vs abandoned:

community actors range from excellent to abandoned. always check last-update date and run count before depending on one.

reliability and uptime

bright data and oxylabs both publish 99.9%+ uptime sla on enterprise plans. real-world data backs this.

apify is at 99.7-99.8% on their compute platform. occasional dataset api hiccups. for non-critical scraping this is fine, for time-sensitive feeds you’d want fallback paths.

if your scraper is mission-critical (real-time pricing for trading, fraud detection feed), bright data or oxylabs are safer. for everything else, apify is plenty.

frequently asked questions

what’s the cheapest of the three for a hobbyist?

apify, by a wide margin. the $5 free monthly credit covers light hobby use. bright data and oxylabs are designed for paying customers.

can i use bright data’s proxies inside apify actors?

yes. apify actors support custom proxy configs. many people use apify for the compute + scheduler + storage and bright data for the proxies.

is apify safer than running scrapers myself?

operationally, yes. they handle retries, error logging, scheduled runs, dataset storage, and monitoring. legally, no. you’re still responsible for what you scrape and how you use the data.

which platform is best for scraping linkedin?

apify has actively maintained linkedin actors. bright data has linkedin in their dataset catalog. oxylabs supports linkedin via the web scraper api but with stricter usage limits. apify is the most flexible.

do any of them solve captchas natively?

bright data’s web unlocker and oxylabs’ web scraper api solve common captchas as part of the managed flow. apify actors integrate captcha solvers via configurable inputs but you bring your own solver budget.

what about smaller competitors like scrapfly or zyte?

different category but overlapping. zyte and scrapfly compete more directly with bright data web unlocker on the api side. neither has apify’s marketplace or bright data’s dataset catalog. for a focused choice between unblocking apis, see our scraperapi vs zyte article.

final thoughts

if you’re a developer or small team looking to ship custom scrapers fast, apify is your platform. if you’re an enterprise that needs the deepest infrastructure and dataset catalog, bright data is the safe pick. if you want clean apis with predictable enterprise sla and don’t need a marketplace, oxylabs is the cleanest.

most production stacks i’ve seen end up using two of the three. apify for custom one-off scrapers, bright data for the heavy unlocking work or pre-scraped datasets. oxylabs for ecommerce-specific apis where their parsing quality saves engineering time. you don’t have to pick just one.

ScraperAPI vs Zyte vs Bright Data Web Scraper: Which API in 2026?

scraperapi is the cheapest of the three for normal sites. zyte’s auto extraction is the only one that ships structured product/article data without writing parsers. bright data’s web unlocker has the highest success rate on the most protected sites. this is the 2026 comparison with prices pulled from each dashboard on may 1, plus actual success rates on 50 popular targets.

the matrix

prices vary by plan and overage. these are entry-tier figures.

scraperapi: the cheapest workhorse

scraperapi is the simplest of the three. one endpoint, one api key, you pass a target url, you get back html. it handles proxy rotation, retries, and basic captcha solving. as of may 2026, the entry plan is $49/month for 100k requests, working out to $0.49 per 1k.

import requests

resp = requests.get(
    "https://api.scraperapi.com/",
    params={
        "api_key": "YOUR_KEY",
        "url": "https://example.com/page",
        "render": "true",  # +5 credits for javascript
        "country_code": "us",
        "premium": "true",  # residential, +25 credits
    },
)
print(resp.text)

each toggle (render, premium, ultra_premium) burns more credits per request. a basic html fetch is 1 credit. a javascript-rendered residential request is 30+ credits. budgets disappear fast on protected sites.

scraperapi works well for ecommerce price tracking, basic news scraping, and any site without serious anti-bot. it struggles on datadome and perimeterx.

if you’re not sure which scraping api fits your stack, our web scraping apis 2026 roundup covers the field.

zyte api: the structured-data api

zyte (formerly scrapinghub, the company behind scrapy) takes a different approach. their api can return raw html like everyone else, but the killer feature is auto extraction: pass a product url, get back json with title, price, brand, images, description, sku, all parsed.

import requests
from base64 import b64encode

api_key = "YOUR_ZYTE_KEY"
auth = b64encode(f"{api_key}:".encode()).decode()

resp = requests.post(
    "https://api.zyte.com/v1/extract",
    headers={"Authorization": f"Basic {auth}"},
    json={
        "url": "https://www.amazon.com/dp/B0CHX1W1XY",
        "product": True,
    },
)
print(resp.json()["product"])

response includes structured fields like:

{
  "name": "Echo Dot (5th Gen)",
  "price": "49.99",
  "currency": "USD",
  "brand": "Amazon",
  "images": [...],
  "description": "...",
  "sku": "B0CHX1W1XY"
}

this skips a huge engineering tax. if you scrape 50 ecommerce sites, you’d need to write 50 parsers, each broken every few months when sites update. zyte’s auto extraction handles all of them.

cost is more variable than scraperapi. simple html fetches are ~$0.40/1k. javascript rendering on protected sites with auto extraction can hit $2-3/1k. bigger budgets, but you save the parser-writing cost.

zyte’s residential success rates on protected sites are excellent. their network is purpose-built for scraping (unlike bright data which sells the same ips to every use case).

bright data web unlocker: the heavy artillery

bright data’s web unlocker is the premium “always works” option. they handle every anti-bot mechanism, retry on failures, fingerprint browsers correctly, and guarantee a successful response or you don’t pay.

import requests

resp = requests.get(
    "https://api.brightdata.com/dca/trigger",
    headers={"Authorization": "Bearer YOUR_TOKEN"},
    json={
        "url": "https://www.example-protected-site.com/",
        "render": True,
        "country": "us",
    },
)
print(resp.text)

pricing is per successful request, $1.05-3 per 1k depending on volume tier. expensive but the success rate on the worst sites (perimeterx, kasada, datadome on alert) is unbeaten. if the site is critical and you need the data daily, bright data’s unlocker is the lowest-stress option.

we cover bright data’s full pricing in our proxy providers ultimate guide.

success rate on 50 popular sites

we ran 1000 requests per site through each api in april 2026. success = 200 status with valid html or auto-extracted data. results aggregated by site difficulty:

scraperapi loses on the hard tier. zyte holds well across the board. bright data is best on extreme.

cost per successful request flips the table:

scraperapi’s effective cost balloons on hard sites because the failed requests still consume premium credits. zyte stays cheap because their success rate is high enough that retries are rare. bright data is flat at $1.05/1k regardless of difficulty.

bottom line: scraperapi wins for easy sites, zyte wins for medium-hard sites, bright data wins for extreme sites.

javascript rendering compared

all three support js rendering. behavior differs:

scraperapi spins up a headless chrome instance per request. you control rendering with render=true. wait for selector with wait_for_selector. roughly 5-10 second response time on rendered requests.

zyte uses their internal headless browser pool. js execution is included on most plans. response times in the 3-7 second range, faster than scraperapi.

bright data’s web unlocker uses a managed scraping browser. fastest of the three at 2-5 seconds. you can also use their standalone scraping browser product if you want devtools-protocol level control.

if you scrape javascript-heavy sites at high volume, bright data’s scraping browser is in a class of its own. zyte is a strong second. scraperapi is the budget option.

auto extraction quality

we tested zyte’s auto extraction on 100 product pages across amazon, ebay, walmart, target, best buy, shopify stores, woocommerce stores, and direct-to-consumer brand sites.

field accuracy:

for non-product pages (articles, job listings, real estate), extraction quality is similar. saves real engineering hours vs writing custom parsers.

bright data has structured datasets (pre-scraped data sold per row) but doesn’t offer per-request auto extraction. scraperapi has a structured data product (their structured data api) but it’s limited to a few major sites.

who should pick what

(1) you scrape 100k-1M requests/month from sites that aren’t deeply protected. scraperapi. cheapest, simplest.

(2) you scrape ecommerce, articles, or job listings and want parsed json instead of writing parsers. zyte. the auto extraction pays for itself.

(3) you scrape protected sites where failures are unacceptable. bright data web unlocker. flat-rate pricing, near-perfect success.

(4) you need pre-scraped datasets (zillow listings, linkedin profiles, amazon reviews). bright data datasets, not the unlocker.

(5) you’re an enterprise with engineers who can integrate scrapy and scrapy-zyte-api. zyte. their library integration is the cleanest of the three.

developer experience

scraperapi: simplest. one endpoint, query params, html out. great docs. limited monitoring dashboard.

zyte: cleanest python ecosystem. scrapy-zyte-api integrates as a downloader middleware. dashboard shows success rates, retries, and cost per project. excellent documentation.

bright data: most complex. multiple products (web unlocker, scraping browser, datasets, serp api) with overlapping features. dashboard is feature-rich but takes time to learn. documentation is thorough but scattered.

frequently asked questions

is scraperapi still worth it in 2026?

yes for non-protected sites and simple scraping. it’s the cheapest tool that does the basics. for protected sites, your money goes further with zyte or bright data.

what’s the difference between zyte’s residential and datacenter tiers?

residential routes through real residential ips, $0.40-3/1k depending on site difficulty. datacenter is faster and cheaper ($0.10-0.20/1k) but only works on non-protected sites. zyte automatically picks based on the target.

can i use bright data’s web unlocker for real-time scraping?

yes, but it’s not optimized for sub-second latency. response times are 2-5 seconds. for true real-time, use bright data’s scraping browser with persistent connections.

does any of them solve captchas?

scraperapi solves recaptcha v2 and v3 on most sites. zyte solves captchas as part of their managed scraping. bright data’s web unlocker handles all common captchas. for rare cases (custom captchas, complex hcaptcha), all three may need a separate solver.

which is best for serp scraping (google search results)?

bright data has a dedicated serp api at $1.50-3/1k. zyte includes serp via their api at similar pricing. scraperapi has a structured google scraper. bright data wins on accuracy, zyte on simplicity.

can i mix and match?

yes. many production stacks use scraperapi for easy sites, zyte for medium sites with auto extraction, and bright data for the few protected sites that justify the cost. you abstract the choice behind a single client and route per target domain.

final thoughts

there’s no single winner here. scraperapi is the cheapest for simple needs. zyte is the most engineering-friendly with the best auto extraction. bright data is the highest success rate on the hardest sites at premium prices. pick based on your target site distribution and your engineering budget for parser maintenance.

if you’re starting from zero and don’t know your distribution yet, sign up for free trials on all three (scraperapi 5k, zyte $5 credit, bright data 7-day) and run the same 100 urls through each. the data answers the question in an hour.

bright data pricing 2026: residential, isp, mobile – what each plan actually costs

bright data’s 2026 residential proxies start at $5.04 per gb on the pay-as-you-go tier, dropping to $3.50/gb on the $499 growth plan and around $2.00/gb at enterprise scale. isp proxies are $1.50 per ip per month on a static-pricing model. mobile proxies are $9.00 per gb. datacenter proxies are $0.066 per ip per ip-day or $0.50 per gb on the shared pool. there is no truly unlimited plan in 2026, the closest is the enterprise commit which negotiates a fixed monthly rate.

bright data’s pricing page lists rates that look reasonable until you realize most users overshoot their gb estimate by 30-50%. this guide unpacks what each plan actually costs in real-world scraping, where the volume discounts kick in, and how the bright data structure compares to oxylabs, smartproxy, and the cheaper rotating-pool providers.

all numbers below come from the bright data pricing page accessed in may 2026 plus my own enterprise quote conversations earlier this year.

the short version

residential is the volume product. mobile is the premium product. isp is the static-ip product. datacenter is the cheap-and-fast product. each has a different unit (gb vs ip-month vs ip-day) which makes apples-to-apples comparisons hard until you do the conversion.

residential proxies in 2026

bright data’s residential pool is the largest in the industry: 150m+ ips across 195 countries. pricing is per-gb of traffic.

the $5.04 retail rate is the sticker price. effectively no one running production scraping pays that. the $3.50/gb tier kicks in at $499 monthly commit. the $2.00/gb tier requires a sales conversation and a 12-month commit.

what eats your gb budget. each browser-based scrape (playwright, selenium) consumes 1-3mb per page including images, css, and ads. so 1gb runs you 300-1000 pages depending on site weight. raw http scraping (requests, httpx) is much lighter, maybe 50kb per page including json apis. so 1gb runs you 20,000+ light pages.

practical estimate for a typical scraper: budget for 2x what you think you’ll use the first month. real consumption almost always overshoots the model.

isp proxies (static residential)

isp proxies are residential ips that don’t rotate. you rent a fixed ip for a month. these are the right answer for any scraping that needs session continuity (logged-in scraping, account management, ad verification).

bandwidth on isp proxies is unlimited at all tiers, which makes the unit economics very different from rotating residential. for use cases that need 100 stable ips and process tons of bandwidth (account management, e-commerce automation), isp at $150/month for 100 ips and unlimited gb is dramatically cheaper than residential at any equivalent gb level.

geo coverage is narrower than rotating residential. isp ips are mostly us, uk, germany, france, india, australia. countries outside the major commercial hubs are limited.

mobile proxies (3g/4g/5g)

mobile is bright data’s premium tier. ips come from real cell carriers, share carrier-grade nat with thousands of legitimate users, and almost never get blocked by anti-bot systems.

mobile at $9/gb is roughly 2x the residential rate. for tough targets (tiktok, instagram, kasada-protected ecommerce, datadome-protected travel sites) the higher pass-rate justifies the cost. for easy targets, residential is the right pick and you don’t need mobile at all.

dedicated mobile providers like singapore mobile proxy and other smaller specialists often beat bright data’s mobile pricing at the lower-tier commit levels because they specialize in carrier-grade networks. for a 5-10gb/month mobile workload, the smaller providers are usually cheaper.

datacenter proxies

cheap and fast, but easily blocked. datacenter ips come from cloud providers (aws, azure, digitalocean, ovh) and any decent anti-bot system blocklists their ranges.

bright data has two pricing models for datacenter:

shared pool (rotating).
– $0.50 per gb on starter
– $0.40 per gb on growth
– $0.25 per gb at enterprise scale

dedicated ips (you own the ip for the month).
– $0.066 per ip-day ($1.98 per ip per month, full month)
– $0.05 per ip-day at volume

for any scraping target that doesn’t use sophisticated anti-bot, datacenter at $0.40/gb is hard to beat on cost. for anything that does (ecommerce, social, finance), datacenter pass rates are too low to be worth the savings.

unlimited bandwidth plans

bright data does not offer a true unlimited residential plan in 2026. the closest is the enterprise commit which negotiates a fixed monthly rate at high volume.

isp proxies and mobile proxies have unlimited bandwidth at all tiers, but you pay per ip not per gb.

if you specifically need unlimited residential, smaller providers like ipfoxy, livinatic, and proxyempire offer “unlimited” plans starting around $79/month with usage caps that effectively limit you to 200-500gb depending on the provider. the unlimited bandwidth rotating proxies guide covers which of those are real and which are theatrically limited.

bright data add-ons

a few non-proxy services that bundle into the platform:

• web unlocker: $5 per 1,000 successful requests. handles cloudflare, kasada, akamai automatically. cheap for low volume, expensive at scale.
• scraping browser: $8.40 per gb for cloud-hosted playwright with built-in anti-bot bypass. niche but useful.
• serp api: $1.50 per 1,000 google searches. competitive with serpapi and dataforseo.
• dataset marketplace: pre-collected datasets (linkedin, amazon, indeed) at custom pricing.

these all share the same wallet so you can mix proxy bandwidth with web unlocker calls in one bill.

how bright data pricing compares

bright data is the most expensive at sticker rates among the major providers. the value differentiator is pool size, geo coverage, and reliability. if you need 195-country coverage, no one else in this list comes close.

for a fuller competitive picture see the proxy provider comparison and the proxy pricing comparison.

hidden costs and gotchas

things that don’t appear on the pricing page but matter for your real bill.

bandwidth multiplier on rendered pages. browser scraping uses 30-50x more bandwidth than raw http. budget accordingly.

failed requests still count. bright data charges for traffic, not for successful scrapes. if your target blocks you and you retry, both attempts count.

geo-targeting premiums. some country/city targets cost a small premium (5-15% over base rate). especially common for hard-to-source geos like china, russia, indonesia.

zone setup time. new zones (custom configurations) sometimes take 24 hours to provision. plan ahead for production launches.

24/7 support is included in $499+ tiers. pay-as-you-go users get email support only, with 24-48 hour response times. on a tight production schedule this matters.

who should pick bright data vs alternatives

pick bright data if you need the largest pool, the broadest geo coverage, the most polished dashboard, or you’re at enterprise scale and need spp-style negotiation. for compliance-sensitive use cases (financial services scraping, regulated industries), bright data’s compliance posture is the strongest in the market.

pick oxylabs if you want a slightly cheaper alternative with comparable polish. oxylabs and bright data are functionally close enough that the choice is usually about pricing on your specific volume.

pick smartproxy or proxyempire if you’re in the $50-500/month sweet spot and you want the best price-per-gb at smaller scales. these are excellent for solo developers and small teams.

pick a specialist mobile provider if mobile is your main need and bright data’s mobile gb cost feels expensive at your volume.

faq

what is the cheapest bright data plan in 2026?
the pay-as-you-go residential at $5.04/gb has no monthly minimum. effective entry cost is $0 if you only scrape occasionally. for serious volume, the $499/month starter at $3.50/gb is the first real tier.

does bright data offer a free trial?
yes. 7-day free trial with $5 in credits for new accounts. enough to test the platform but not enough for any real production work.

how much does bright data cost per month?
depends entirely on volume. typical small scraper bill: $50-200/mo on pay-as-you-go residential. typical mid-sized: $499-2000/mo on growth or business tiers. enterprise: $5000-50000/mo.

is bright data unlimited?
no, on residential and mobile. yes, on isp and on dedicated datacenter ips (per-ip pricing with no bandwidth cap).

can i get a refund from bright data?
yes. unused balance on prepaid plans is refundable for 30 days after the last activity. enterprise contracts have specific refund clauses negotiated case-by-case.

how does bright data compare to oxylabs?
both are tier-1 enterprise providers. oxylabs is roughly 10-15% cheaper at most tiers. bright data has the larger pool. for a deep dive see the oxylabs vs bright data comparison.

conclusion

bright data’s pricing in 2026 is premium. you pay 10-30% above the closest competitor in exchange for the biggest pool, the most polished tooling, and enterprise-grade compliance.

for occasional scraping, pay-as-you-go residential at $5.04/gb is fine. for any serious workload, the math forces you toward the $499+ tiers where the real per-gb rates kick in. for unlimited bandwidth, isp proxies are the right answer at $1.50/ip/month with no gb cap.

map your real volume estimate (and double it for safety) before committing. the difference between the $499 and $999 tiers can be more than the difference between bright data and a smaller competitor for the same workload.