String vs Oxylabs (2026): 100 sites compared

Where the 28-point success gap appears

Oxylabs returned nothing on all five attempts for 24 sites: Temu, Mouser, Trustpilot, Apartments.com, Realtor.com, Skyscanner, G2, StubHub, Ticketmaster, SeatGeek, StockX, Zara, Kayak, Allegro, Zoopla, Congress.gov, Facebook, Amazon, Booking.com, AT&T, CarGurus, AXS, Roblox and Stack Overflow.

String returned content on 23 of those 24 and passed all five attempts on 22, with Allegro at four of five. Temu shut out both providers. Oxylabs was partial on another 14 sites; String was partial on five.

The Amazon result needs context. The benchmark used the general Web Unblocker path. Oxylabs separately publishes dedicated Amazon scrapers with parsers for products, search results, pricing, sellers, and bestsellers. The 0-for-5 benchmark result describes Web Unblocker, not the dedicated Amazon product. If you want parsed Amazon rows from String instead, the same engine ships as the Amazon Product and Amazon Search Apify Actors, and our guide to how to scrape Amazon carries the Python and TypeScript we ran against product and search pages.

The same caution applies to every buyer. Test the actual source and options you plan to purchase using a subset of the public harness.

Anti-bot systems and categories

Oxylabs led String on no anti-bot system in this run. Its closest columns were PerimeterX and Kasada; its weakest were AWS WAF and Cloudflare. Its clearest win was a single site: Idealista, five of five attempts where String returned nothing.

String's larger advantage appeared by business category. Oxylabs passed none of the ticketing requests, 53.3% of review sites, and 57.1% of real-estate requests. String passed 100%, 96.7% and 85.7%. Oxylabs' residential proxy FAQ also says some categories, including ticketing, government, streaming, banking, and mailing, are restricted on that proxy product, and Oxylabs returned nothing on Congress.gov, the benchmark's one government site. The benchmark tested Web Unblocker, not the residential proxy product, so the published proxy restriction is not a proven cause of a Web Unblocker result.

What each product is

Oxylabs Web Unblocker, the product the benchmark ran, is a proxy, not a REST API. You send your normal request through unblock.oxylabs.io:60000, and Oxylabs handles proxy selection, browser fingerprints, retries, and optional JavaScript rendering. Options travel as request headers. Oxylabs also sells a Web API whose /scrape endpoint accepts a URL and can add geolocation, JavaScript rendering, browser instructions, and parsing. Its dedicated scrapers cover Amazon, Google, YouTube, and other common targets.

The platform goes beyond fetching. Oxylabs provides browser instructions, a custom parser, real-time and asynchronous integrations, and plan-specific rate limits. It also sells residential proxies and other proxy types for teams that want direct session and routing control.

String is narrower. Send a URL to the Web Access API and get Markdown, HTML, or JSON. Proxy rotation, fingerprinting, CAPTCHA solving, and browser escalation happen behind the endpoint. String also provides search, an asynchronous sitemap crawl, remote Chrome over CDP, and an MCP server. It does not provide raw proxy access or Oxylabs' catalog of dedicated parsers.

Both providers publish agent integrations. Oxylabs maintains an official MCP server. String's MCP server is listed as ai.usestring/web-access and exposes fetch, search, and sitemap tools.

Oxylabs also has the stronger procurement package. Its Trust Center lists ISO/IEC 27001:2022, SOC 2 Type 1 and Type 2, and SOC 3 materials. If one of those controls is a buying gate, it may outweigh the benchmark difference.

The same request in Oxylabs and String

One job, done both ways: fetch a JavaScript-heavy product page and get its content back. Each request follows the vendor's own docs for that job, checked against those docs on October 28, 2026. Each one reads its key from an environment variable, such as $STRING_API_KEY.

curl -k -x https://unblock.oxylabs.io:60000 \
  -U "$OXYLABS_USERNAME:$OXYLABS_PASSWORD" \
  -H "X-Oxylabs-Geo-Location: United States" \
  -H "X-Oxylabs-Render: html" \
  "https://www.example.com/product/123"

curl https://request.usestring.ai/v1/fetch \
  -H "Authorization: Bearer $STRING_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"url": "https://www.example.com/product/123", "format": "markdown"}'

This is the Web Unblocker request the benchmark sent, plus one render header. The target URL stays the request URL, and the username and password go to the proxy. The docs ask you to ignore the proxy's SSL certificate, so -k is required. X-Oxylabs-Render: html loads the page in a browser, and the docs ask for a 180-second client timeout when you use it. The benchmark adapter did not send the render header. Oxylabs' Web Unblocker docs suggest it "if you are observing low success rates or retrieve empty content," so a run with it could score higher. Oxylabs' separate Web API takes a bearer key at webapi.oxylabs.io/v1/scrape and uses run_js: true instead. String takes the key in a header, picks the fetch path and proxy class for each request, and bills the path it used. Pages that need JavaScript are rendered without executeJS.

Pricing: gigabytes and counted results are not verified pages

Oxylabs now sells one subscription whose monthly amount can be spent on any of its products. Its Web Unblocker pricing, for the product the benchmark ran, lists these self-service rates:

Web Unblocker bills traffic, not pages. Its billing documentation says you are billed for "both request and response traffic of successfully extracted data." It defines success more broadly than verified content: 2xx and 4xx target responses count as successful, even when the result does not contain the expected information. Oxylabs does not bill its own 5xx errors. That means a target's 403 block page can count against the plan even though the benchmark treats it as a failure.

Page weight sets the cost of each page. At the 201 kB per request in Oxylabs' own billing example, Basic's 16 GB covers about 79,600 requests, or about $1.26 per 1,000. At 500 kB per request it covers about 32,000 requests, or about $3.13 per 1,000. Both are our arithmetic, not Oxylabs' figures. The docs also say JavaScript rendering can make the traffic for a job larger than the HTML file alone.

The same subscription also buys Oxylabs' Web API /scrape endpoint, which bills per result and which the benchmark did not test. On Basic it lists most sites at $0.90 per 1,000 results without JavaScript and $1.40 with it, Amazon at $0.45, and Google without JavaScript at $0.90.

String's plan fee is paid in as usage credit. String bills a request when it returns the site's response, including a site's 404, and does not bill blocked requests. The first 20,000 standard fetches are free with no card. Starter is $19 per month and includes $19 of usage; Growth is $99 and includes $99. Usage above that bills at the same rates, and unused credit rolls over. Growth rates are:

At the $100 level, String's standard request, premium request, and browser on a standard proxy all cost less than Web Unblocker's derived $1.26 per 1,000 at 201 kB a page. String's $3.25 browser-plus-premium path costs more than both derived Web Unblocker figures. Against the per-result Web API, String is cheaper on standard requests and on browsers through a standard proxy, and the two tie at $0.90 on a premium request against a non-rendered result. The Web API's $1.40 rendered rate is cheaper than String's $3.25 path, and its $0.45 Amazon rate is for a dedicated scraper String does not provide. The target mix and the page weight determine the answer.

Do not divide the list rate by benchmark success for either product. Their billing rules are not the same as the benchmark's content-marker rule. Success rate answers whether the page arrives. Billing documentation answers what counts against the plan. Keep those questions separate.

Oxylabs' raw proxy product is billed by the gigabyte too. On the same subscription, residential proxies cost $6 per GB on Starter, up to 5 GB, $5 per GB on Basic, up to 20 GB, and $2.50 per GB from 1,000 GB on Enterprise. Headers, response size, assets, and retries all consume bandwidth. A gigabyte cannot be compared with a verified page by looking at one headline price.

Where Oxylabs is the better choice

• Idealista, AutoTrader or Foot Locker matters more than the overall benchmark.
• You need raw residential, ISP, mobile, or datacenter proxies with direct session control.
• You want dedicated Amazon parsers, custom parsers, browser instructions, or a dashboard playground.
• Oxylabs' ISO and SOC materials satisfy a procurement requirement.
• Your own test shows a dedicated Web API scraper returning the full dataset at its lower per-result rate.

Where String is the better choice

• A fixed list of protected pages must return consistently through one interface.
• Your targets resemble the benchmark's ticketing, reviews, real estate, retail, or travel shutouts.
• You want Markdown output without maintaining parsers or proxy sessions.
• Plain fetches make up most of the workload, where String Growth lists $0.16 per 1,000 successful requests.
• You do not want a 403 block page to count as billable traffic.

Using String and Oxylabs together

Keep Oxylabs for raw proxy workloads and dedicated sources such as Amazon, then send the protected URLs that miss on Web Unblocker through String. The reverse can also work: use String as the default fetch layer and Oxylabs when a workflow needs dedicated parsing, browser instructions, or direct proxy control.

Switching from Oxylabs to String

What carries over. A Web Unblocker GET moves from a proxy setting to an API call. The URL goes in the body of a POST to String's /v1/fetch, and the username and password become a bearer key. X-Oxylabs-Geo-Location becomes countryCode (a two-letter code). You can drop X-Oxylabs-Render, because String renders pages that need it without being asked. A Web Scraper API realtime POST with source: "universal" maps the same way, with geo_location to countryCode and browser instructions to String's actions.

What you rebuild. Web Unblocker sessions, dedicated sources and parsers such as amazon_product and google_search, the Custom Parser, Push-Pull async jobs, batches of up to 5,000 jobs, the Scheduler, cloud-storage delivery, and the residential proxy pool. String's jsonSchema covers parsing for the fields you name and /search covers live Google results. Amazon jobs on a dedicated source often stay with Oxylabs.

Integrations to reconnect. Oxylabs publishes an MCP server for its newer Web API, a LangChain package, a LlamaIndex integration, agent skills, and a Postman collection. Its docs list no n8n, Zapier, or Make integration. String ships no language SDK, so an SDK call becomes one HTTP POST. Its hosted MCP server runs at mcp.usestring.ai, n8n calls String through the HTTP Request node, and Zapier, LangChain, and the Vercel AI SDK have setup guides in String's docs.

Test on your own failing URLs first. Take the URLs that miss on Oxylabs today and run them through String, with the same Oxylabs configuration beside them. Compare pages returned and cost per usable page, then move only the targets that improve. Leave custom headers off for the test: String does not bill a blocked request unless that request sent custom headers. The first 20,000 String standard fetches are free with no card. If Oxylabs is one option among several, the Oxylabs alternatives comparison ranks the substitutes on the same benchmark.

FAQ

Is String better than Oxylabs for protected websites?

On the September 2026 Web Data Frontier Benchmark overall, yes. String returned verified content on 97.0% of 500 requests and Oxylabs on 69.0%, with 2 total shutouts against 24. Oxylabs returned Idealista where String failed. Run the public test harness against your own target list.

Is String cheaper than Oxylabs?

Sometimes. Oxylabs Basic is $100 for up to 16 GB of Web Unblocker traffic at $6 per GB, about $1.26 per 1,000 requests at 201 kB a page by our arithmetic. String Growth's standard request costs $0.16 per 1,000 successful requests and its premium request $0.90, both below that figure. String's $3.25 browser-plus-premium path costs more. Oxylabs' per-result Web API lists $0.90 for most non-rendered results and $1.40 with JavaScript, with lower rates for dedicated Amazon and Google sources.

Does Oxylabs charge for blocked requests?

Its Web Unblocker billing documentation says 2xx and 4xx target responses count as successful results, while its own 5xx errors are not billed. A target 403 can therefore count against the plan even when it contains a block page. String does not bill blocked requests. It bills a site's 404, because that is the site's real answer.

Which product has the larger proxy network?

Oxylabs. Its residential proxy page advertises more than 175 million IPs, and it also sells mobile, ISP, and datacenter products. String does not sell raw proxies.

Which product is faster?

String, and the gap holds on pages both products returned. On the 59 sites where String and Oxylabs each returned all five attempts in the September 2026 run, String's median was 2.50 seconds against Oxylabs' 5.10, and String was quicker on 46 of the 59. The failure-aware benchmark score points the same way, 7.06 seconds against 15.86, but that score also charges each provider for the sites it fails, so it is not a raw response-time figure.

When should I stay with Oxylabs?

Stay when Idealista or a dedicated Oxylabs source is central to the job, when you need raw proxy control or dedicated Amazon and Google parsers, or when Oxylabs' enterprise compliance package is a procurement requirement. The benchmark tested the general Web Unblocker path and does not measure every dedicated source.

Can I verify these results?

Yes. The benchmark repository, Oxylabs adapter, and September results file are public. A full run makes thousands of billable requests, so begin with a subset.