{"id":402,"date":"2026-04-23T21:14:00","date_gmt":"2026-04-23T18:14:00","guid":{"rendered":"https:\/\/proxy-man.com\/blog\/?p=402"},"modified":"2026-04-23T21:14:00","modified_gmt":"2026-04-23T18:14:00","slug":"datacenter-residential-mobile-three-proxies-three-completely-different-jobs-mixing-them-up-costs-real-money","status":"publish","type":"post","link":"https:\/\/proxy-man.com\/blog\/datacenter-residential-mobile-three-proxies-three-completely-different-jobs-mixing-them-up-costs-real-money\/","title":{"rendered":"Datacenter, Residential, Mobile: Three Proxies, Three Completely Different Jobs. Mixing Them Up Costs Real Money"},"content":{"rendered":"<header><em>&#8220;Proxy&#8221; is a category word the way &#8220;vehicle&#8221; is a category word. A bicycle, a semi-truck, and an ambulance are all vehicles \u2014 but if you order a bicycle for a hospital run, someone dies. The proxy market is functionally split into three radically different infrastructures with roughly a 10\u201350x cost spread per GB and a 10x difference in detection resistance. Most of the money lost in this industry is not lost to bad providers. It is lost to teams buying a panel van when they needed a semi \u2014 or a semi when they needed a bicycle.<\/em><\/header>\n<section id=\"section-1\" aria-labelledby=\"heading-1\">\n<h2><\/h2>\n<h2 id=\"heading-1\">The $2.8B Word That Means Three Different Things<\/h2>\n<p>The global proxy service market was valued at approximately\u00a0<strong>$2.8 billion in 2023<\/strong>, growing at roughly 20% CAGR \u2014 driven almost entirely by the anti-bot arms race between scraping teams on one side and Cloudflare, Akamai, and PerimeterX on the other. Cloudflare alone sits in front of approximately 20% of all websites on the public internet. Any serious data-collection or multi-account operation is effectively negotiating with one anti-bot vendor&#8217;s ruleset at massive scale.<\/p>\n<p>Underneath that single market number sit three products that share almost nothing except a protocol:<\/p>\n<ul>\n<li><strong>Datacenter proxies<\/strong>\u00a0\u2014 $0.50\u2013$2 per IP\/month, sub-50ms latency, unlimited bandwidth, ASN tagged as hosting.<\/li>\n<li><strong>Residential proxies<\/strong>\u00a0\u2014 $2\u2013$15 per GB, 100\u2013500ms latency, bandwidth is the constraint, ASN tagged as consumer ISP.<\/li>\n<li><strong>Mobile proxies<\/strong>\u00a0\u2014 $30\u2013$150\/month per channel or premium per-GB, 4G speeds, CGNAT-shared IPs, ASN tagged as mobile carrier.<\/li>\n<\/ul>\n<p>Each type optimizes a different triangle:\u00a0<strong>(cost per GB) \u00d7 (detection resistance) \u00d7 (speed)<\/strong>. No provider can max all three \u2014 physics and economics forbid it. The rest of this piece is a practical taxonomy: what each type actually is, where it dominates, where it fails, and how to choose before touching a pricing page.<\/p>\n<aside><strong>Sources:<\/strong>\u00a0Grand View Research, Proxy Service Market Report, 2023. W3Techs, Usage statistics of Cloudflare, 2024.<\/aside>\n<\/section>\n<section id=\"section-2\" aria-labelledby=\"heading-2\">\n<h2><\/h2>\n<h2 id=\"heading-2\">Datacenter Proxies: The 18-Wheeler. Cheap per Mile, Useless in a Back Alley<\/h2>\n<p>A datacenter proxy is an IP address hosted in a commercial data center \u2014 AWS, Hetzner, OVH, DigitalOcean \u2014 with an ASN that any anti-bot system recognizes as belonging to a hosting company, not a human being. This is not hidden information. MaxMind, IPinfo, and IP2Location tag these ranges continuously, and Cloudflare Bot Management, Akamai Bot Manager, and Kasada all maintain ASN reputation lists updated in near-real-time. An IP range flagged on one customer&#8217;s site propagates across the network within minutes.<\/p>\n<p><strong>Where datacenter proxies still dominate:<\/strong>\u00a0bulk scraping of targets with no serious anti-bot \u2014 open government data portals, small e-commerce with no Cloudflare, Google Shopping feed collection at low volume, SERP scraping against non-FAANG engines, internal QA smoke tests, and any workload where speed and cost per GB matter more than ASN classification.<\/p>\n<p><strong>Where they fail immediately:<\/strong>\u00a0any site running Cloudflare Enterprise, PerimeterX, Kasada, or Meta&#8217;s internal anti-bot system has the entire Hetzner\/OVH\/AWS\/M247 ASN range effectively blocklisted. Open-source scraping benchmarks from communities like Apify and ScrapingBee show datacenter proxy success rates on Cloudflare-protected sites dropping\u00a0<strong>below 10%<\/strong>\u00a0as of 2024, versus 70\u201390%+ on unprotected targets.<\/p>\n<p>The myth worth busting: &#8220;premium&#8221; datacenter proxies don&#8217;t escape this. Cloudflare does not care whether your hosting bill is $5 or $500 per month. It cares about the ASN. Nike&#8217;s Akamai configuration blocks the entire Hetzner\/OVH\/AWS range during sneaker drops, regardless of which &#8220;exclusive&#8221; datacenter provider sold you the IP.<\/p>\n<aside><strong>Sources:<\/strong> Akamai Bot Manager product documentation. Cloudflare Bot Management whitepaper. ScrapingBee, &#8220;Proxy Success Rate Benchmarks,&#8221; 2024 blog series.<\/aside>\n<aside><\/aside>\n<aside><\/aside>\n<\/section>\n<section id=\"section-3\" aria-labelledby=\"heading-3\">\n<h2 id=\"heading-3\">Residential Proxies: The Rental Car with Local Plates. Slower, Pricier, but It Drives Through the Gated Neighborhood<\/h2>\n<p>A residential proxy uses an IP address that belongs to a real consumer ISP \u2014 Comcast, AT&amp;T, Deutsche Telekom, Verizon \u2014 routed through an end-user device that has joined a peer-to-peer network, typically via an SDK embedded in a free app or service. To MaxMind, to Netflix, to Cloudflare&#8217;s bot scoring engine, this IP is indistinguishable from a normal home user \u2014 because it is a normal home IP.<\/p>\n<p><strong>Where residential proxies dominate:<\/strong>\u00a0web scraping behind Cloudflare, multi-accounting on social platforms, ad verification, price intelligence on major marketplaces like Amazon and Walmart, geo-specific content testing, and streaming geo-checks. Success rates on Cloudflare-protected targets that reject datacenter IPs outright typically run 70\u201385% with a well-rotated residential pool.<\/p>\n<p><strong>The hidden cost trap<\/strong>\u00a0that burns teams who do not calculate before buying: residential is billed per GB, and modern websites are heavy. The 2024 HTTP Archive Web Almanac puts the average page weight on top-1,000 e-commerce sites at approximately\u00a0<strong>2.7MB on desktop<\/strong>. A scraping job of 500,000 product pages requires over 1TB of residential traffic \u2014 at $10\/GB, that is $10,000 in bandwidth before a single row lands in your dataset. Teams that start with residential because &#8220;it just works&#8221; and do not benchmark first routinely discover a four-figure bill in four hours.<\/p>\n<p><strong>The ethical conflict that the industry pretends does not exist:<\/strong>\u00a0residential proxy pools are built on consent mechanisms buried in the terms of service of free apps. The 2019 controversy around Hola VPN and Luminati \u2014 now Bright Data \u2014 established publicly that millions of users were routing commercial traffic through their devices with no clear awareness. In 2023, the FTC settled with BIScience over exactly this model applied to free browser extensions. The sourcing question is not paranoia. It determines whether the user whose device you&#8217;re routing through consented in any meaningful sense. Ask every provider where their IPs come from before you sign a contract.<\/p>\n<aside><strong>Sources:<\/strong>\u00a0Trend Micro research, &#8220;Hola VPN and Luminati Proxy,&#8221; 2015\u20132019. FTC press release, BIScience settlement, January 2023. HTTP Archive, Web Almanac 2024, Page Weight chapter.<\/aside>\n<\/section>\n<section id=\"section-4\" aria-labelledby=\"heading-4\">\n<h2><\/h2>\n<h2 id=\"heading-4\">Mobile Proxies: The Diplomat&#8217;s Car. You Can&#8217;t Stop It at Any Checkpoint<\/h2>\n<p>A mobile proxy uses an IP address assigned by a mobile carrier to a 4G\/LTE device \u2014 T-Mobile, AT&amp;T Mobility, Verizon Wireless. What makes these IPs structurally different from every other proxy type is\u00a0<strong>CGNAT: Carrier-Grade NAT<\/strong>. Mobile carriers share a single public IPv4 address among hundreds to thousands of concurrent subscribers. RFC 6598 reserves the 100.64.0.0\/10 range specifically for this shared-address infrastructure.<\/p>\n<p>The practical consequence: a platform that aggressively bans a mobile IP bans a neighborhood&#8217;s worth of legitimate users simultaneously. Meta, TikTok, Google, and Instagram are all aware of this asymmetry \u2014 their abuse-detection systems avoid hard blocks on mobile carrier ASNs because the collateral damage is too high. This structural immunity is why mobile proxies carry the highest trust score across every major anti-bot system.<\/p>\n<p><strong>Where mobile proxies dominate:<\/strong>\u00a0high-value, high-risk account operations \u2014 ad agency Meta Business Manager at scale, TikTok Shop, Instagram multi-account client work, sneaker drops on mobile-first retailers, account warmup sequences, and any platform where a residential IP still gets flagged due to behavioral profiling.<\/p>\n<p><strong>Economics:<\/strong>\u00a0$30\u2013$150\/month per dedicated channel, or premium per-GB pricing. This is not markup \u2014 it is the cost floor set by real SIM cards, real 4G hardware, and real carrier contracts. There is no software shortcut that produces a genuine mobile ASN.<\/p>\n<p><strong>The myth worth busting:<\/strong>\u00a0mobile proxies are not slow. Modern 4G delivers 20\u201350Mbps, which is faster than many cheap residential setups. The speed constraint is latency variability from the radio layer, not throughput.<\/p>\n<p><strong>The uncomfortable supply-chain issue:<\/strong>\u00a0SIM farms in low-cost jurisdictions are regularly marketed as &#8220;US mobile&#8221; or &#8220;EU mobile&#8221; proxies. The ASN fingerprint reveals the carrier&#8217;s actual country of registration. Verify the carrier&#8217;s ASN against a public BGP toolkit before committing to a mobile provider \u2014 a $40\/month &#8220;US mobile&#8221; proxy routing through a Southeast Asian carrier does not give you a US mobile IP score.<\/p>\n<aside><strong>Sources:<\/strong>\u00a0RFC 6598, IANA-Reserved IPv4 Prefix for Shared Address Space, 2012. Meta Engineering blog, &#8220;Fighting abuse at scale&#8221; series.<\/aside>\n<\/section>\n<section id=\"section-5\" aria-labelledby=\"heading-5\">\n<h2><\/h2>\n<h2 id=\"heading-5\">The Cost\/Risk Matrix: A One-Page Decision Framework<\/h2>\n<p>The decision depends on two variables: the anti-bot sophistication of your target, and your required volume. Map your workload against both axes before touching a pricing page.<\/p>\n<table>\n<caption>Proxy type selection matrix by target anti-bot tier and job volume<\/caption>\n<thead>\n<tr>\n<th scope=\"col\">Target anti-bot tier<\/th>\n<th scope=\"col\">Low volume (&lt;100K req\/day)<\/th>\n<th scope=\"col\">Medium volume (100K\u201310M)<\/th>\n<th scope=\"col\">High volume (&gt;10M)<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td><strong>None \/ minimal<\/strong>\u00a0(open data, small sites)<\/td>\n<td>Datacenter \u2014 rotating<\/td>\n<td>Datacenter \u2014 rotating<\/td>\n<td>Datacenter \u2014 bulk pool<\/td>\n<\/tr>\n<tr>\n<td><strong>Cloudflare Free \/ basic WAF<\/strong><\/td>\n<td>Residential \u2014 rotating<\/td>\n<td>Residential \u2014 rotating<\/td>\n<td>Residential + cost audit first<\/td>\n<\/tr>\n<tr>\n<td><strong>Cloudflare Enterprise \/ Akamai \/ PerimeterX<\/strong><\/td>\n<td>Residential \u2014 sticky sessions<\/td>\n<td>Residential \u2014 sticky + behavioral delay<\/td>\n<td>Residential + engineering review<\/td>\n<\/tr>\n<tr>\n<td><strong>Meta \/ TikTok \/ Google \/ Instagram<\/strong><\/td>\n<td>Mobile \u2014 sticky<\/td>\n<td>Mobile \u2014 dedicated channel<\/td>\n<td>Mobile \u2014 dedicated + account warmup<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p><strong>The counterintuitive math on cost per successful request<\/strong>\u00a0\u2014 which is the only number that actually matters:<\/p>\n<p>Scenario: you need 1 million completed product pages from an Amazon category. Average page weight: 3MB. You are comparing a datacenter proxy at $1\/GB versus a residential proxy at $10\/GB.<\/p>\n<ul>\n<li>Datacenter at 8% success rate on this target: 12.5M attempts needed to land 1M completed pages. At 3MB per attempt: ~37.5TB of traffic. At $1\/GB:\u00a0<strong>$37,500<\/strong>\u00a0\u2014 plus engineering time for retries.<\/li>\n<li>Residential at 80% success rate: 1.25M attempts. At 3MB per attempt: ~3.75TB. At $10\/GB:\u00a0<strong>$37,500<\/strong>\u00a0\u2014 and you finish the job in a fraction of the time.<\/li>\n<\/ul>\n<p>In this example the cost is the same \u2014 but the residential job completes. The datacenter job generates 11.5 million failed requests, burns engineering hours on retry logic, and likely triggers progressive rate limiting that degrades success rate further over time. The Google SRE book describes this class of error precisely:\u00a0<em>optimizing for the cheapest attempt is a local maximum that produces a globally terrible system.<\/em><\/p>\n<p><strong>The right reframe is not &#8220;which proxy is cheaper.&#8221; It is &#8220;which proxy is cheaper per completed row in the final dataset.&#8221;<\/strong><\/p>\n<aside><strong>Sources:<\/strong>\u00a0Betsy Beyer et al.,\u00a0<cite>Site Reliability Engineering<\/cite>, O&#8217;Reilly, 2016 (chapters on SLOs and error budgets). HTTP Archive Web Almanac 2024.<\/aside>\n<\/section>\n<section id=\"section-6\" aria-labelledby=\"heading-6\">\n<h2><\/h2>\n<h2 id=\"heading-6\">The Mistake Nobody Wants to Admit: Buying from Three Providers When You Need One<\/h2>\n<p>The operational default for any team that has grown past basic scraping is three separate vendor relationships: one datacenter provider for bulk, one residential provider for Cloudflare targets, one mobile provider for account work. This arrangement is not a sign of sophistication. It is a sign that the tooling market has not caught up to how teams actually work.<\/p>\n<p>The operational tax of three vendors: three dashboards, three invoicing cycles, three API schemas to maintain in your codebase, three support queues, three prepaid balances sitting idle when one job type is quiet. When a provider gets blocked or goes down \u2014 and they do \u2014 you are migrating under time pressure with a live workload.<\/p>\n<p>The natural next step in this market is consolidation: one balance, one API, one support channel, the ability to switch proxy type as the job demands without switching vendors or rewriting integration code. That is the architecture PROXY-MAN is built around \u2014 datacenter, residential, and mobile under one dashboard, REST API and Telegram bot, HTTP(S)\/SOCKS5, sticky and rotating sessions, and \u2014 because it sits inside the SMS-MAN ecosystem \u2014 the same balance as virtual phone numbers for teams that need both proxy and SMS verification in the same workflow.<\/p>\n<p>This is not a model for everyone. It is specifically useful for teams that have already lived through at least one of the failure modes described in this piece: the surprise bandwidth bill, the sneaker drop that failed on every datacenter IP, the agency meeting three weeks before 80 accounts need to go live. Those teams already know they need all three proxy classes. The question is whether they want three vendors or one.<\/p>\n<p>The proxy infrastructure market is currently where CDN was in 2012: fragmented specialist vendors, buyer personas still maturing, consolidation 10\u201315 years behind what Cloudflare eventually did for the CDN\/WAF\/bot management stack. The teams that consolidate early carry less operational overhead and smaller stale-balance write-offs at year end.<\/p>\n<aside><strong>Source:<\/strong>\u00a0Cloudflare investor relations materials, 2019\u20132024. Ben Thompson \/ Stratechery analysis of infrastructure aggregation.<\/aside>\n<\/section>\n<section id=\"section-7\" aria-labelledby=\"heading-7\">\n<h2><\/h2>\n<h2 id=\"heading-7\">A 15-Minute Self-Audit Before You Renew Any Proxy Contract<\/h2>\n<p>Answer these seven questions about your actual workload before you open a pricing page. Forward this list to your team. The answers determine the proxy class \u2014 not the provider&#8217;s marketing.<\/p>\n<dl>\n<dt><strong>1. What is the anti-bot tier of your primary target?<\/strong><\/dt>\n<dd>Check whether the target runs Cloudflare, Akamai, PerimeterX, or Kasada. A quick\u00a0<code>curl -I<\/code>\u00a0on the domain and a look at response headers will tell you in 30 seconds. If it&#8217;s Cloudflare Enterprise or Meta\/Google, datacenter proxies will not complete the job regardless of price.<\/dd>\n<dt><strong>2. What is your realistic monthly GB volume at 2.5MB average page weight?<\/strong><\/dt>\n<dd>Multiply your target request count by your average page weight. If the answer is in the terabytes, run the cost-per-successful-request math from Section 5 before committing to residential. A bandwidth estimate that feels fine at $1\/GB becomes a different conversation at $10\/GB.<\/dd>\n<dt><strong>3. What is your tolerance for IP rotation within a session?<\/strong><\/dt>\n<dd>Rotating proxies are cheaper and sufficient for stateless scraping. Sticky sessions \u2014 same IP for the duration of a session \u2014 are mandatory for anything that involves login, cart, or multi-step flows. Confirm the provider offers sticky sessions at the tier you need before signing.<\/dd>\n<dt><strong>4. Do you need geo-targeting at country, state, or city precision \u2014 and can you verify the provider&#8217;s claims?<\/strong><\/dt>\n<dd>Country-level geo is standard. City-level geo on residential or mobile is a premium feature and often overpromised. Verify claimed geo with an independent IP lookup tool (ipinfo.io, BGP toolkit) on a sample of IPs before committing to a contract that depends on precision geo.<\/dd>\n<dt><strong>5. What is your cost per successful outcome, not per attempt?<\/strong><\/dt>\n<dd>Run the math in Section 5 against your specific success rate assumptions. If you do not know your current success rate by proxy class and target, your current proxy spend is probably wrong.<\/dd>\n<dt><strong>6. Do you need API integration for CI\/CD, or is a dashboard enough?<\/strong><\/dt>\n<dd>If your proxy usage is baked into automated pipelines, an API with documented authentication and session management is non-negotiable. A dashboard-only provider is a manual dependency in an automated workflow \u2014 a failure point at 3 AM.<\/dd>\n<dt><strong>7. If this provider gets breached, delisted, or shuts down, what is your migration cost?<\/strong><\/dt>\n<dd>This is the quietly devastating question. If the answer is &#8220;significant re-engineering and days of downtime,&#8221; you are carrying vendor-concentration risk you may not have priced. The answer tells you whether a specialist provider or a platform vendor is the lower-risk choice for your workload.<\/dd>\n<\/dl>\n<p>Question 7 is the one most teams skip. It is also the one that most reliably predicts whether a proxy relationship becomes a liability during a live incident.<\/p>\n<\/section>\n<footer>\n<section aria-label=\"Conclusion\"><em>Every industry has a word that hides three things underneath. In freight, &#8220;truck&#8221; means pickup, panel van, or semi. In finance, &#8220;bond&#8221; means treasury, corporate, or junk. In proxies, the word is &#8220;proxy&#8221; \u2014 and the three things underneath have a 50x cost spread and radically different use cases.<\/em><em>Most of the money lost in this industry is not lost to bad providers. It is lost to teams buying a panel van when they needed a semi, or a semi when they needed a bicycle. The burned scraping budget, the sneaker drop that failed on every datacenter IP, the agency meeting three weeks before 80 accounts need to go live \u2014 these are not edge cases. They are the modal failure mode.<\/em><em>The fix is not buying more expensive proxies. The fix is matching the tool to the job, and having a platform that lets you switch between tools without switching vendors. The proxy market will consolidate in the next 3\u20135 years the same way CDN did. The teams that stop juggling three providers now are the ones with the smaller stale-balance write-offs and the bigger completed-dataset count at the end of the quarter.<\/em><\/section>\n<section aria-label=\"Call to action\"><\/section>\n<section id=\"faq\" aria-labelledby=\"faq-heading\">\n<h2 id=\"faq-heading\">Frequently Asked Questions<\/h2>\n<dl>\n<dt><strong>What is the difference between datacenter, residential, and mobile proxies?<\/strong><\/dt>\n<dt><\/dt>\n<dd>Datacenter proxies use IPs from hosting providers \u2014 cheap and fast but flagged by anti-bot systems. Residential proxies use real home ISP IPs \u2014 expensive per GB but pass most geo-checks. Mobile proxies use carrier IPs with CGNAT \u2014 the most expensive and most trusted tier, used for high-risk accounts on Meta, TikTok, and similar platforms.<\/dd>\n<dt><strong>When should I use a datacenter proxy vs a residential proxy?<\/strong><\/dt>\n<dt><\/dt>\n<dd>Use datacenter proxies when your target has no serious anti-bot protection. Switch to residential when the target runs Cloudflare, Akamai, or PerimeterX \u2014 datacenter ASNs are blocklisted by those systems at the network level, and no &#8220;premium&#8221; datacenter product escapes the ASN check.<\/dd>\n<dt><strong>Why are mobile proxies more expensive than residential proxies?<\/strong><\/dt>\n<dd>Mobile proxies require real SIM cards, real 4G hardware, and real carrier contracts. The cost floor is set by physical infrastructure. Any &#8220;mobile proxy&#8221; priced at residential rates is not running on genuine carrier ASNs \u2014 verify with an independent IP lookup tool before buying.<\/dd>\n<dt><strong>Do residential proxies work with Netflix and streaming services?<\/strong><\/dt>\n<dd>Residential proxies use IPs from real consumer ISP ASNs, which streaming services cannot distinguish from regular home users. This is why they bypass geo-blocks that VPNs and datacenter proxies fail \u2014 the ASN check returns &#8220;residential&#8221; rather than &#8220;hosting.&#8221;<\/dd>\n<dt><strong>What is the cheapest proxy per successful request, not per GB?<\/strong><\/dt>\n<dd>It depends on your target&#8217;s anti-bot tier. A $1\/GB datacenter proxy at 8% success rate on a Cloudflare-protected site costs more per completed request than a $10\/GB residential proxy at 80% success \u2014 because failed attempts still consume bandwidth, engineering time, and re-run cost. Always calculate cost per successful outcome.<\/dd>\n<\/dl>\n<\/section>\n<\/footer>\n","protected":false},"excerpt":{"rendered":"<p>&#8220;Proxy&#8221; is a category word the way &#8220;vehicle&#8221; is a category word. A bicycle, a semi-truck, and an ambulance are all vehicles \u2014 but if&hellip;<\/p>\n","protected":false},"author":4,"featured_media":405,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-402","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-proxy-definition"],"aioseo_notices":[],"_links":{"self":[{"href":"https:\/\/proxy-man.com\/blog\/wp-json\/wp\/v2\/posts\/402","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/proxy-man.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/proxy-man.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/proxy-man.com\/blog\/wp-json\/wp\/v2\/users\/4"}],"replies":[{"embeddable":true,"href":"https:\/\/proxy-man.com\/blog\/wp-json\/wp\/v2\/comments?post=402"}],"version-history":[{"count":3,"href":"https:\/\/proxy-man.com\/blog\/wp-json\/wp\/v2\/posts\/402\/revisions"}],"predecessor-version":[{"id":406,"href":"https:\/\/proxy-man.com\/blog\/wp-json\/wp\/v2\/posts\/402\/revisions\/406"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/proxy-man.com\/blog\/wp-json\/wp\/v2\/media\/405"}],"wp:attachment":[{"href":"https:\/\/proxy-man.com\/blog\/wp-json\/wp\/v2\/media?parent=402"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/proxy-man.com\/blog\/wp-json\/wp\/v2\/categories?post=402"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/proxy-man.com\/blog\/wp-json\/wp\/v2\/tags?post=402"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}