Independent · measured · dated
Best proxies for web scraping, tested on 9 real sites
Every provider on this page gets the same requests to Amazon, Google, Zillow, Walmart and five more, from the same server, on the same day. What came back is the ranking. No stars, no sponsored placement.
Latest run
- Tested
- 10 Sep 2026
- Targets
- 9real sites
- Providers
- 3+ datacenter baseline
- Baseline pass rate
- 15 / 45no proxy
Quick answer · 10 Sep 2026
The best proxies for web scraping in our September 2026 test, ranked by how many of 45 identical requests to 9 real sites brought back the real page:
- Oxylabs 20/45 passed · median 1.8 s · from $6/GB Best for the highest pass rate on protected retail and real-estate sites.
- Decodo 19/45 passed · median 2.1 s · from $2/GB Best for the lowest price per successful request and a self-serve start.
- SOAX 19/45 passed · median 2.1 s · from $5/GB Best for sticky sessions of custom length and the cheapest way to try residential proxies.
- Bright Data test pending · from $8.4/GB Best for enterprise pipelines that need ZIP-level targeting and a compliance team.
Baseline without a proxy, from the same datacenter server: 15/45 passed, median 0.3 s. Residential proxies bought access to Zillow; the 5 sites that blocked everything (Google, Walmart, Best Buy, LinkedIn, Indeed) block the missing browser, not the address; the browser run below shows which of them a real browser opens.
Residential proxies · rotating · US
Pass rate against 9 targets, 5 requests each
Success means the real page arrived, not a 200 with an empty shell. Every request goes out from a different exit IP. Time is the median for successful requests. Prices are list prices on the test date.
| Provider | Pass rate | Median time | IP pool | Geo-targeting | From | |
|---|---|---|---|---|---|---|
| No proxyDatacenter IP · the baseline every provider must beat | 0.34 s | 1 | — | $0 | ||
 Best in testResidential · 100M+ IPs · 195 countries | 1.82 s | 100M+ | City · ASN | $6/GB | See Oxylabs pricing | |
 Test pendingResidential · 72M+ IPs · 195 countries | Test pending | – | 72M+ | City · ASN · ZIP | $8.4/GB | See Bright Data pricing |
 Residential · 115M+ IPs · 195+ locations | 2.05 s | 115M+ | City · ZIP · ASN | $2/GB | See Decodo pricing | |
 Residential · 155M+ IPs · 195+ locations | 2.09 s | 155M+ | City · ISP | $5/GB | See SOAX pricing |
We earn a commission if you buy through the pricing links. It does not change the numbers: the test script, the target list and the raw results are public, and providers do not see results before you do.
Target by target
Which sites let which proxy through
Five requests per site, browser headers, no retries, no JavaScript. Green: all passed. Amber: some. Red: none. The reason shown is what the site sent back. Four targets have their own page: Zillow, where a residential proxy works and a headless browser does not; Indeed, which opens only for both together; Amazon, which needed neither; and LinkedIn, which refused everything.
| Target | No proxy | Oxylabs | Bright Data | Decodo | SOAX |
|---|---|---|---|---|---|
| api.ipify.orgour IP (control) | 5/5 | 5/5 | – | 5/5 | 5/5 |
| Amazonsearch results | 5/5 | 5/5 | – | 5/5 | 4/5 |
| Googlesearch results | 0/5page came back | 0/5page came back | – | 0/5Google "sorry" block page | 0/5page came back |
| Zillowlistings | 0/5HTTP 403 | 5/5 | – | 4/5 | 5/5 |
| Walmartsearch results | 0/5redirected to a /blocked page | 0/5redirected to a /blocked page | – | 0/5timed out with no response (30 s) | 0/5redirected to a /blocked page |
| Best Buyhomepage | 0/5connection dropped mid-response (curl 92) | 0/5connection dropped mid-response (curl 92) | – | 0/5connection dropped mid-response (curl 92) | 0/5connection dropped mid-response (curl 92) |
| LinkedIncompany page | 0/5HTTP 999 | 0/5HTTP 404 | – | 0/5HTTP 404 | 0/5HTTP 404 |
| Indeedjob search | 0/5HTTP 403 | 0/5HTTP 403 | – | 0/5HTTP 403 | 0/5HTTP 403 |
| Nikecategory page | 5/5 | 5/5 | – | 5/5 | 5/5 |
Same proxies, real browser
What changes when a headless Chromium does the requests
The same nine targets, the same five requests each, the same exit networks, but this time a real Chromium loads the page, runs its JavaScript and waits up to 15 seconds for the content or for a challenge to resolve. No stealth plugins, no CAPTCHA solving. Images, fonts and media are not loaded; HTML, CSS, scripts and API calls are.
| Provider | Plain requests (curl) | Headless Chromium | Browser gained | Browser lost | Median (browser) | Traffic (browser) |
|---|---|---|---|---|---|---|
| No proxy (datacenter) | Best Buy 5/5 | – | 1.4 s | 31 MBvs 2 MB plain | ||
| Indeed 4/5 | Zillow 0/5 | 3.6 s | 27 MBvs 3 MB plain | ||
| Indeed 4/5 | Zillow 0/5 | 3.6 s | 21 MBvs 3 MB plain | ||
| Indeed 5/5 | Zillow 0/5 | 4.3 s | 23 MBvs 3 MB plain |
Two things to take from this table. A residential address plus a real browser opens a Cloudflare-protected site (Indeed) that neither alone could; that is the case for pairing the two. And a plain headless browser is a fingerprint of its own: Zillow, which let plain requests through from every residential provider, refused every headless Chromium request from the same networks. Traffic per request is seven to ten times higher with a browser, which changes the cost per successful request accordingly.
Providers
Four providers, one test
Best in test- City and ASN targeting in every plan
- Rotating and sticky sessions
- Starter 5 GB, cancel any time
Test pending- City, ASN and ZIP targeting
- Pay as you go, no commitment
- Scraping browser and dataset marketplace
Tested 10 Sep 2026- City, ZIP and ASN targeting included
- Pay as you go $4/GB
- 3-day free trial, 100 MB
Tested 10 Sep 2026- Country, city and ISP targeting
- Sticky sessions with custom length
- 3-day trial, 400 MB, $1.99
Method
How we test
The same script, the same day, the same server. Anything else would be an opinion.
The nine targets are Amazon search results, Google search results, Zillow listings, Walmart search, the Best Buy homepage, a LinkedIn company page, an Indeed job search, a Nike category page and api.ipify.org as the control that proves the proxy is in the path. Each request carries the headers of a current desktop Chrome but no JavaScript engine, which is how most scrapers built on curl, Python requests or httpx look to a target. A response counts as a pass only when it contains a marker that exists on the real page and nowhere else: a product grid, a listing card, a result count. Timeouts, connection resets, redirects to a block page and full-size pages with empty result containers all count as failures, and the reason is recorded. Exit addresses are looked up per request so that rotation and the owning network are facts, not claims from a sales page. The raw results of every run are published as JSON: openwebdata.io/wp-content/uploads/owd/runs/.
Provider by provider
What each proxy did on the nine targets
Ranked by requests passed. The numbers are from the run above; the text is what the numbers mean for a buyer.
1. Oxylabs 20/45 passed
Best for: the highest pass rate on protected retail and real-estate sites.
- Passed
- api.ipify.org, Amazon, Zillow, Nike
- Blocked
- Google, Walmart, Best Buy, LinkedIn, Indeed
- Median time
- 1.82 s
- Exit addresses
- 5/5 unique · AS7922 Comcast Cable Communications, LLC
- List price
- from $6/GB · Starter 5 GB, $30/mo, cancel any time
- Trial
- None on self-service
- Targeting
- City · ASN · Rotating and sticky
Oxylabs took the top spot by a single request, and the request that made the difference was on Zillow: five out of five listings pages came back with real content, from five different Comcast addresses. Amazon search and the Nike category page also passed every time. Where Oxylabs failed, everyone failed: Google returned a full-size page with no results in it, Walmart redirected to its blocked page, Best Buy cut the connection mid-transfer, LinkedIn and Indeed refused outright.
Median time to a successful response was the fastest of the three residential providers, about six times slower than the direct datacenter request, which is normal for a residential hop. Pricing starts higher than Decodo or SOAX, and there is no free trial on the self-service plan, but city and ASN targeting come with every plan and the 5 GB starter can be cancelled at any time.
2. Decodo 19/45 passed
Best for: the lowest price per successful request and a self-serve start.
- Passed
- api.ipify.org, Amazon, Zillow 4/5, Nike
- Blocked
- Google, Walmart, Best Buy, LinkedIn, Indeed
- Median time
- 2.05 s
- Exit addresses
- 5/5 unique · AS20115 Charter Communications LLC
- List price
- from $2/GB · Pay as you go $4/GB, plans from $2/GB
- Trial
- 3-day trial, 100 MB
- Targeting
- City · ZIP · ASN · Rotating and sticky (minutes to days)
Decodo, the provider formerly known as Smartproxy, finished one request behind Oxylabs. Amazon and Nike passed five out of five; Zillow let four of five through, the one miss being a PerimeterX challenge on a Charter Communications address. Google was the only target where Decodo behaved differently from the others: instead of an empty page it got Google's explicit "sorry" rate-limit page, which is at least an honest failure. Walmart timed out after 30 seconds rather than redirecting.
At the list price it is the cheapest of the tested providers, and the plan price is a third of Oxylabs. With 19 of 45 requests passing, the price per successful request is the lowest in this run. City, ZIP and ASN targeting are included, sticky sessions can last from minutes to days, and the three-day trial with 100 MB is enough to reproduce this test yourself.
3. SOAX 19/45 passed
Best for: sticky sessions of custom length and the cheapest way to try residential proxies.
- Passed
- api.ipify.org, Amazon 4/5, Zillow, Nike
- Blocked
- Google, Walmart, Best Buy, LinkedIn, Indeed
- Median time
- 2.09 s
- Exit addresses
- 5/5 unique · AS20115 Charter Communications LLC
- List price
- from $5/GB · Pay as you go $5/GB, $3/GB on $200/mo plan
- Trial
- 3-day trial, 400 MB, $1.99
- Targeting
- City · ISP · Rotating and sticky, custom length
SOAX tied with Decodo on total requests passed but got there differently: Zillow passed five out of five, Amazon four out of five, Nike five out of five. Like Oxylabs, it received a full-size Google page with no results in it and was redirected to Walmart's blocked page. Exit addresses rotated on every request, all of them Charter Communications on this run, which matches what the target sites saw.
SOAX is the only provider of the three with a paid trial cheap enough to be a rounding error: 400 MB for $1.99 over three days, which is a hundred times the traffic this test used. Pay as you go is priced between Decodo and Oxylabs, and the $200 monthly plan brings it close to Decodo. Sessions can be pinned for a custom length and targeting goes down to city and ISP, useful when a site behaves differently per carrier.
4. Bright Data Test pending
Best for: enterprise pipelines that need ZIP-level targeting and a compliance team.
- List price
- from $8.4/GB · Pay as you go, no commitment
- Trial
- Free trial for verified businesses
- Targeting
- City · ASN · ZIP · Rotating and sticky
Bright Data is the provider every other comparison page ranks first, and it is the one we have not been able to measure yet. Residential access requires a business verification (KYC) before the account is activated; ours is in review, and the test runs the day it clears. Until then the row says "Test pending" rather than a number we did not measure.
What is known without a test: the pay-as-you-go list price is the highest of the four, the targeting options are the widest (city, ASN and ZIP code), and the product range goes well beyond proxies into a scraping browser and ready-made datasets. That breadth is what enterprise buyers pay for; whether it buys a higher pass rate on the nine targets is exactly what the next run will show.
Proxy types
Which proxy type do you actually need?
The test above is residential. The baseline row is datacenter. Here is what the difference bought, and where the other two types fit.
Addresses owned by hosting companies. Fastest and cheapest by a wide margin, and in this run they were enough for Amazon search and the Nike category page. Blocked outright by Zillow. Baseline: 15/45 · 0.3 s
Real consumer addresses from ISPs such as Comcast and Charter, rotated per request. The only type that got into Zillow. Roughly six times slower than datacenter and priced per gigabyte. We also measured how the rotation itself behaves: 100 requests, how many addresses, how many networks. Best in test: 20/45 · 1.8 s
Datacenter-hosted addresses registered under a consumer ISP. Keep the same address for hours or days, which logins and carts need. Priced per address, not per gigabyte. Not in this run; planned for the next.
Addresses behind a carrier’s shared NAT, so blocking one blocks thousands of real phones. Hardest to block and the most expensive per gigabyte. Worth it only for mobile-first targets that reject everything else.
On the radar
Providers we have not tested yet
Listed because buyers ask about them. They get a row in the table the month they are measured, not before.
- Bright Data — the most-cited provider on every other comparison page. Residential access needs business verification; ours is in review and the test runs the day it clears. Listed above with its list price and “Test pending”.
- IPRoyal — pay-as-you-go residential traffic that does not expire, popular with small scrapers. Next in line.
- Webshare — the cheapest well-known datacenter and residential option, with a free tier. Useful as the low-price reference point.
- DataImpulse — residential from $1 per gigabyte, the lowest list price we know of. Cheap traffic is only cheap if it gets in, which is what the test is for.
- NetNut — do not buy. As of 9 September 2026 the netnut.io domain resolves to a seizure notice from the US Federal Bureau of Investigation, and its name servers point to fbi.seized.gov. Comparison pages that still rank it first have not been updated. We removed it from this page the day we noticed.
Guides
What we learned doing it
Baseline · 11 Sep
What happens when you scrape without a proxy
Nine sites, one datacenter IP. Five blocked it outright. Google sent an empty page with status 200.
Reference · 14 Sep
curl with a proxy, the complete reference
Every flag that matters, with real output: auth, HTTPS tunnelling, the socks5 vs socks5h DNS leak, timing with -w.
Test · 16 Sep
We tested 300 free proxies to see what they leak
13 % were alive. One in four of those handed our real IP to the target. Two of them were somebody’s home router.
FAQ
Before you buy
Which proxies are best for web scraping?
On 10 Sep 2026, measured against 9 real sites with 45 requests per provider, Oxylabs (20/45) passed the most requests, followed by Decodo (19/45), SOAX (19/45). The differences are small and come almost entirely from Zillow. Pick by the target matrix for the sites you scrape, then by price per gigabyte.
Why is "no proxy" in the table?
Because it is the baseline. If a provider does not beat a plain datacenter address, it is not worth paying for. In this run a direct request from our server passed 15 of 45; residential proxies added Zillow. Amazon and Nike let the datacenter address in anyway.
Why do all providers score so close to each other?
Because on a single plain HTTP request the target decides by address type and by whether a real browser is present, and all tested providers deliver real consumer addresses (Comcast and Charter on this run). The sites that blocked everything (Google, Walmart, Best Buy, LinkedIn, Indeed) block the missing browser, not the IP. The browser run on this page shows what changed when the same proxies were driven by a real Chromium: some of those sites opened, and one that plain requests passed closed.
How much does a residential proxy cost for scraping?
List prices of the providers on this page run from $2 to $8.4 per gigabyte. Traffic is what you pay for: this test moved about 3.0 MB per 45 requests, roughly 67 MB per 1,000 page requests, which is $0.13 to $0.56 per 1,000 requests at list price. Pages with images or JavaScript rendering cost several times more.
Can residential proxies be detected?
Yes, and this test shows how. Walmart, Best Buy, LinkedIn and Indeed blocked every request from every provider even though each came from a real Comcast or Charter address. Modern bot defences look at the TLS fingerprint, headers and JavaScript execution before they look at the IP. A residential address removes one signal; it does not make a script look like a browser.
Residential or datacenter proxies for web scraping?
Datacenter addresses are faster (median 0.3 s here versus about 2 s residential) and far cheaper, and they were enough for Amazon search and the Nike category page. Residential addresses were the only way in to Zillow. Start with datacenter, move a target to residential only when the datacenter address is blocked.
Is web scraping illegal?
Not in itself. Collecting publicly available data is lawful in most jurisdictions, and US courts have held that accessing public pages is not "unauthorised access" under the CFAA. What can be unlawful is what you collect and what you do with it: personal data, copyrighted content, bypassing a login, or breaching a contract you agreed to. This page is not legal advice; check the target site's terms and your local law.
Do providers pay for placement?
No. We earn a commission on purchases through the pricing links, and that is disclosed above the results. Providers do not see results before publication and cannot buy a position. The test script, target list and raw results are published.
How often is this page updated?
The test is rerun on the same day each month from the same server, and old results stay published with their date. The date at the top of the results table is the date of the numbers you are reading. Prices are checked on the same day.