13 Proxy Scrapers & Free Proxy Lists (2026)

A proxy scraper is a program that walks public proxy listings — forums, aggregator sites, GitHub repositories, JSON feeds — and pulls out the address, port and protocol of every proxy it finds. The good ones then try each proxy against a real URL and throw away whatever fails, which is the step that decides whether the resulting list is useful or just long.

Below are thirteen options: five open-source tools you run yourself, four desktop or hosted products, and four maintained lists that let you skip the scraping entirely. Each entry states the same five attributes, so you can compare them without reading every paragraph. Everything here was checked against the vendor's own pricing page, documentation or repository on 11 August 2026, and where a project has gone quiet or a claim could not be confirmed, the entry says so rather than glossing over it.

What a proxy scraper does, and what it does not do

A scraper collects candidates. A checker decides which candidates work. Most tools worth using do both, but the distinction matters, because an unchecked list of ten thousand proxies is close to worthless: the addresses come from misconfigured servers and open relays that appear and vanish within hours. When a tool advertises a huge number, that number is almost always candidates found, not proxies that work right now.

What a proxy scraper cannot do is create capacity. It finds proxies other people are running, usually without those people intending it. That has consequences for reliability, for speed, and for what the operator of that proxy can see — covered further down under the real cost of free proxies.

The thirteen tools and sources at a glance

Open-source tools you run yourself

These are free, auditable, and give you control over which sources get scraped and how aggressively each proxy is verified. They also mean you own the traffic: checking a large list opens hundreds of simultaneous connections to untrusted hosts, which some ISPs and most consumer routers dislike.

1. proxy-scraper-checker

• What it is: A scraper and checker in one, with an interactive terminal UI

• Runs on: Windows, Linux, macOS and Android as a single binary; Docker image available

• Protocols: HTTP, SOCKS4, SOCKS5

• Cost: Free, MIT licensed

• Watch out for: No GUI beyond the terminal, and geolocation fields stay empty unless your check URL returns the exit IP

The most complete free option at the moment. It pattern-matches proxies out of raw text, HTML or JSON, so you can point it at almost any source without writing a parser, and it accepts scheme://user:pass@host:port as well as CIDR ranges from URLs or local files. Every proxy has to fetch a real URL in full to survive, which drops the ones that connect and then stall — a failure mode most checkers miss. Output is JSON with response time, exit IP, ASN and city-level geolocation resolved from offline MaxMind databases, plus plain text files per protocol. Note that this project is now written in Rust, not Python; older roundups (including an earlier version of this page) still describe it as a Python script.

2. ProxyBroker2

• What it is: An async finder, checker and rotating proxy server

• Runs on: Python 3.10 to 3.14, or the published Docker image

• Protocols: HTTP(S), SOCKS4, SOCKS5, and CONNECT to ports 80 and 25

• Cost: Free, Apache-2.0 licensed

• Watch out for: The current release is a 2.0.0 beta, and the original ProxyBroker it forked has not moved meaningfully in years

ProxyBroker2 is the maintained continuation of ProxyBroker, and it is the one to reach for if you want more than a list. Its four subcommands are find (collect and check), grab (collect without checking), serve (run a local proxy server that rotates through what it found) and update-geo. That serve mode is the differentiator: your scraper points at one local address and never has to think about rotation. It draws on around 50 sources, filters by protocol, country, response time and anonymity level, checks DNSBL status, and verifies that each proxy passes cookies and referrers.

3. ProxyPool

• What it is: A self-hosted proxy pool with a scheduler and an HTTP API

• Runs on: Python with Redis; official Docker image and Compose file

• Protocols: HTTP and HTTPS

• Cost: Free, MIT licensed

• Watch out for: The README and documentation are largely in Chinese, and there is no SOCKS support

ProxyPool is the most widely used project in this space by a wide margin, and it is built around a different idea: rather than producing a file, it runs two processes — a scheduler that collects and re-validates continuously, and a web service that hands proxies out. Your scraper calls /get for a random proxy, /pop to take one out of the pool, /all for everything, /count for the size, and /delete when a proxy turns out to be dead. Adding your own source means dropping one file into fetcher/sources/ that yields host:port. If you want a pool that behaves like infrastructure rather than a download, this is the shape you want.

4. free-proxy

• What it is: A small Python library that returns a working free proxy on request

• Runs on: Any Python environment — pip install free-proxy

• Protocols: HTTP and HTTPS

• Cost: Free, MIT licensed

• Watch out for: It reads the free-proxy-list family of sites, so it inherits their coverage and their outages

The lowest-effort option if you just need a proxy inside a script you are already writing. free-proxy exposes a single class you construct with filters — country, anonymity, HTTPS only, timeout — and call to get an address back, optionally after it has been tested. It is not trying to be a pool or a pipeline, and that is the point. Version 1.2.2 landed on PyPI in July 2026, so it is being kept current.

5. mubeng

• What it is: A checker and rotating proxy server. It does not scrape

• Runs on: A single Go binary for Linux, macOS and Windows

• Protocols: HTTP(S), SOCKS4, SOCKS5

• Cost: Free, Apache-2.0 licensed

• Watch out for: You have to bring your own list, and the last push was October 2025

Included because the tools above answer only half the problem. mubeng takes a list you already have, checks it quickly, and then exposes a single local endpoint that rotates through the survivors on every request or every N requests. Pair it with any of the list sources further down and you have a working rotating proxy in two commands, without writing rotation logic into your scraper.

Desktop and hosted tools

6. GSA Proxy Scraper

• What it is: A Windows harvester with a port scanner, filters and a built-in proxy server

• Runs on: Windows 7 through 11 and Windows Server; version 4.07, June 2026

• Protocols: Web (HTTP/HTTPS), CONNECT, SOCKS4 and SOCKS5

• Cost: One-time licence with lifetime updates — the vendor checkout quoted CHF 78.41 excluding VAT, about USD 97; a 5-day trial with limited sources is available

• Watch out for: Windows only, and the feature surface is large enough to be genuinely confusing at first

The only paid desktop tool here that has clearly been maintained for years. Beyond harvesting, GSA Proxy Scraper runs an internal proxy server on a local port so other applications can route through its verified pool with rotation handled for them, filters by country, speed and anonymity level, includes a multi-threaded port scanner for finding proxies that are not on any list, and has a small scripting language for sources that need cookies or a login. It also markets a "Google-passed" filter that tests whether a proxy can still query Google — useful if that is your target, irrelevant otherwise. The licence is genuinely one-time; there is no subscription.

7. Somiibo Proxy Scraper

• What it is: A proxy-scraping module inside the Somiibo desktop automation app

• Runs on: Windows 10+, macOS 10.15+ (Intel and Apple Silicon), Ubuntu 18.04+

• Protocols: HTTP, SOCKS4, SOCKS5

• Cost: Free to download and use, with a paid premium tier for heavier use

• Watch out for: No public changelog or version history, so you cannot tell how recently it was updated; it is a module in a broader automation product rather than a standalone tool

Somiibo's proxy scraper is the most approachable graphical option, and unlike GSA it runs on macOS and Linux as well. It scrapes and tests in parallel, filters by protocol, country and anonymity level, and keeps its own source list updated so you are not maintaining one. The honest caveat is verifiability: the product page describes the feature set clearly but publishes nothing about release cadence, so treat "actively maintained" as unconfirmed.

8. Proxy List Scraper (Chrome extension)

• What it is: A browser extension that extracts a proxy table from the page you are on

• Runs on: Chrome and Chromium-based browsers

• Protocols: Whatever the page you are viewing publishes

• Cost: Free

• Watch out for: Version 0.0.4, last updated 17 December 2023, roughly 1,000 users — treat it as abandonware

Worth knowing about mainly so you can judge it accurately. Proxy List Scraper does one small thing: you open a site that lists proxies, click the extension, and it pulls the addresses into a copyable list. That saves a minute of manual selection and nothing more — it does not verify anything, does not aggregate sources, and breaks whenever a source site changes its markup. Given it has not been updated since late 2023, assume some sources no longer parse.

9. Apify Ultimate Proxy Scraper

• What it is: A hosted Actor on the Apify platform that sweeps more than 30 proxy sources

• Runs on: Apify's cloud, callable from its API, JS/Python clients, CLI or MCP

• Protocols: HTTP, HTTPS, SOCKS4, SOCKS5

• Cost: USD 5 per month rental plus Apify platform usage, after a trial period

• Watch out for: Its Apify Store page currently reports a 0% run success rate and one monthly user — verify it runs before relying on it

The pitch is reasonable: no installation, JSON in and a dataset out, and it slots into a pipeline you already run on Apify. Output per proxy is minimal — host, port, combined address and type — with a maxItems cap and a required proxy configuration for its own requests. The problem is the reliability signal. As of 11 August 2026 the Actor store listing shows runs succeeding 0% of the time and a single monthly user, alongside a 4.95-star rating earned when it presumably worked. That is not a claim that it is dead, but it is a strong argument for spending the trial period testing it before paying the rental.

Lists and feeds, so you can skip the scraping

If what you actually want is a list rather than a tool, someone is already running a scraper for you and publishing the output. These are free, need no signup, and update far more often than you would bother to.

10. monosans/proxy-list

• What it is: A free proxy list re-checked every hour, with per-proxy metadata

• Runs on: Nothing — plain files served by GitHub

• Protocols: HTTP, SOCKS4, SOCKS5

• Cost: Free, MIT licensed, no signup, no key, no rate limit

• Watch out for: No HTTPS-specific or CONNECT breakdown, and no country filter — you filter the JSON yourself

The best-documented free list of the group, and the one to try first. It is generated by running proxy-scraper-checker on a schedule, so every entry had to fetch a real URL in full within the last hour. You get proxies/all.txt with protocol prefixes, one bare host:port file per protocol, and a proxies.json carrying response time, exit IP, ASN and city for each proxy. Entries are sorted fastest first, so the top of each file is the best of what was working at the last update, and every commit message records the per-protocol counts.

11. Proxifly's free proxy list

• What it is: A free list validated every five minutes, distributed several ways

• Runs on: jsDelivr CDN URLs, an npm module, or a pip package

• Protocols: HTTP, HTTPS, SOCKS4, SOCKS5

• Cost: Free, GPL-3.0 licensed, no key

• Watch out for: The GPL licence matters if you redistribute the data; the company also sells a paid rotating proxy API

The fastest refresh cycle of anything here. Proxifly's list is re-validated every five minutes and published as JSON, plain text and CSV, split by protocol, deduplicated, with country data attached. At the update immediately before this page was checked it held 2,543 working proxies across 107 countries — a useful number to anchor expectations on, since it is what a continuous web-wide sweep actually yields at any given moment.

12. TheSpeedX/PROXY-List

• What it is: A large, long-running raw proxy list

• Runs on: Nothing — plain text files on GitHub

• Protocols: HTTP, SOCKS4, SOCKS5

• Cost: Free, no key

• Watch out for: No response times, no geolocation, no stated validation method, and the author explicitly disclaims responsibility for the proxies

The volume option: 7,318 proxies at the last update, refreshed roughly daily. Useful when you want maximum candidates and intend to check them yourself with one of the tools above — which is the right way to use it, because nothing in the repository tells you how or when each proxy was verified. One quirk to know: the download links point at the author's sibling SOCKS-List repository rather than the one you are reading.

13. free-proxy-list.net

• What it is: A long-standing HTML table of free HTTP(S) proxies

• Runs on: A browser, or your own parser

• Protocols: HTTP and HTTPS only — SOCKS lives on the operator’s sibling sites

• Cost: The table is free; the API and country filtering belong to the operator’s paid list, priced from USD 59.95 per month

• Watch out for: There is no free API. Roundups that credit this site with one are describing the paid product

Free Proxy List is the site most people mean when they say "free proxy list", and it still works: the page states a ten-minute refresh, and entries in the table were seconds old when checked, with anonymity level, country, HTTPS support and a Google-passed column. What it does not offer is programmatic access. The API URL, custom output formats, country filtering and the 30-minute update cadence advertised further down the page are features of the operator's paid HTTP proxy list, which starts at USD 59.95 per month. For the free table you parse HTML, or use the free-proxy library above, which does exactly that.

How to choose

• You want a usable list in the next minute: take monosans/proxy-list or Proxifly. Both are checked more recently than anything you could scrape yourself in that time.

• You want control over sources and verification: proxy-scraper-checker. It is the only free tool here that insists each proxy complete a real request before it counts.

• You want your scraper to call one endpoint and forget about rotation: ProxyPool if you are happy self-hosting Redis, ProxyBroker2's serve mode if you want it in one command, mubeng if you already have a list.

• You want a proxy inside a Python script and nothing more: the free-proxy library.

• You want a GUI on Windows and will pay once: GSA Proxy Scraper. On macOS or Linux, Somiibo is the graphical option.

• You want it hosted: Apify's Actor is the only one of its kind here, but check its current run success rate before you commit.

What free proxies actually cost

Every tool on this page finds the same underlying thing: servers that are relaying traffic for strangers, usually because someone misconfigured them. That origin, not the software, is what determines what you get.

• The operator sees your traffic. Plain HTTP through a proxy is readable and modifiable by whoever runs it. HTTPS protects the payload, but the operator still learns which hostnames you are visiting and when.

• They disappear. A proxy that passed a check an hour ago is frequently gone now, which is why the useful lists re-verify hourly or faster and why unchecked lists mislead.

• They are often already blocked. Public proxy addresses circulate widely, so the sites you most want to reach have usually seen them.

• Speed is unpredictable. You are sharing an unknown amount of someone else's bandwidth with an unknown number of other users.

• Coverage is whatever turns up. You cannot ask a scraped list for a residential IP in a specific city; you take what the sweep found. Anonymity levels vary just as arbitrarily.

• Nobody is accountable. There is no support channel, and no consent trail behind the IPs you are routing through.

None of that makes free proxies useless. For learning how IP rotation behaves, for one-off checks of how a page renders from another country, or for load that you can afford to have fail, a scraped and freshly checked list is a reasonable tool.

When to stop scraping proxies

The honest dividing line is whether a failed request costs you anything. Once a job has a deadline, a data-quality requirement, or a target that actively defends itself, time spent maintaining a free pool exceeds what paid capacity costs, and the failure modes stop being amusing.

At that point you are shopping for a provider rather than a scraper, and the questions change: how the IPs were sourced, what the success rate is against your specific targets, whether you can hold a session, and whether anyone answers when something breaks. We build Evomi’s residential, mobile, datacenter and static ISP networks around exactly those questions, and residential, datacenter and mobile all come with a free trial so you can compare real success rates against a scraped list before paying anything. If you would rather not manage proxies at all, a scraper API moves that problem entirely off your plate. Either way, benchmark it against your own targets rather than taking our word for it — that is the same advice this page gives about every other tool on it.

How this list was checked

Each entry was verified on 11 August 2026 against the tool's own pricing page, documentation, package registry entry or repository, all linked in the entry itself. Prices are as quoted by the vendor on that date. Where an attribute could not be confirmed from a primary source — Somiibo's release cadence, for instance — the entry says so instead of estimating. Repository activity, licences and version numbers come from GitHub, PyPI and the Chrome Web Store directly.

Four claims in the previous version of this page were wrong or out of date and have been corrected: proxy-scraper-checker is written in Rust rather than Python; a second, similarly named Rust project it was previously listed alongside has since been archived by its author and has been dropped; free-proxy-list.net's API is part of its paid plan, not the free table; and the Chrome extension has not been updated since December 2023.