Proxies for web scraping: residential and datacenter IPs for public data

What is a web scraping proxy?

A web scraping proxy is a server that forwards your scraper's requests to the website and returns the answer. The site sees the proxy's IP address and location instead of yours, so you can collect public pages as they appear in the country you are studying.

A scraping proxy service gives you a pool of these addresses with one login, so your crawler can spread its requests over many IPs and pick the location of each one. Pages that change by country, such as prices, stock levels, search results and ads, then show the version a local visitor would see.

The proxy does not change what you are allowed to collect. It changes where your requests come from. Collect public data, follow each site's rules, and keep your request rate reasonable.

Which proxy should I use for scraping?

Start from the job, not the proxy. The table shows the usual starting point for each job.

Residential packages with city and ISP targeting →Dedicated datacenter IPv4 by country →Static ISP plans for long sessions →

Rotating vs sticky sessions for crawls

On residential proxies, rotation decides how long you keep one IP. Set it on the main login or on each access list in your dashboard, and change it when the job changes.

Per-IP proxies: rotate in your code

Datacenter IPv4, static ISP and IPv6 proxies keep the same address for the whole term; a mobile proxy keeps its host, ports and login, but the carrier IP can change during the term. To spread a crawl, download your order as a .txt list and let your crawler pick a different proxy per request or per site. Scrapy, requests and Playwright all make that a few lines of code.

How many proxies and how much traffic do I need?

Residential proxies are billed per GB of traffic, so the size of each page sets the cost. Datacenter IPv4, static ISP, mobile and IPv6 proxies are billed per IP for a term, with no per-GB charges, so you plan by the number of IPs, not by page size.

Example 1, residential: 2,000 product pages a day at about 250 KB of HTML each is about 500 MB a day, or around 15 GB over 30 days. Packages are valid 30 days, and the price per GB falls as the package gets bigger, down to $1.58 per GB.

Example 2, full browser: 1,000 pages a day loaded in Playwright with images and scripts can reach about 2 MB each, or around 60 GB over 30 days. Load only what you need, such as the HTML or one JSON endpoint, and the same job can use a fraction of that.

Example 3, per IP: 10 datacenter IPv4 proxies cost from $4.50, depending on the country and term, with no per-GB charges. IPv4 terms are 7, 14 or 30 days.

These page sizes are estimates. Run a small test, read the usage in your dashboard (updated every 10 minutes), and size up from real numbers.

Live prices for every type and term →

Set up proxies in curl, Python, Scrapy and Playwright

1. Copy your details. Open My proxies. Datacenter IPv4, static ISP and mobile orders give an IP, an HTTP port, a SOCKS5 port, a login and a password; IPv6 orders have one port for the protocol chosen at checkout. Residential gives one gateway host and port, with a main login and a login for each access list.
2. Use the proxy URL. Write it as http://login:password@host:port, or socks5h://login:password@host:port for SOCKS5. Replace the capitals in the examples below. In Playwright, pass the login and password as separate fields and use the HTTP port: Chromium cannot send a login to a SOCKS5 proxy.
3. Check the exit IP. Request https://t7proxy.com/ip through the proxy. It should print the proxy's IP, not yours.

curl -x http://LOGIN:PASSWORD@HOST:PORT https://t7proxy.com/ip

# pip install requests
import requests

proxy = "http://LOGIN:PASSWORD@HOST:PORT"
r = requests.get("https://t7proxy.com/ip", proxies={"http": proxy, "https": proxy}, timeout=30)
print(r.text)

# settings.py: keep the crawl polite
# ROBOTSTXT_OBEY = True
# AUTOTHROTTLE_ENABLED = True
# DOWNLOAD_DELAY = 1

import scrapy

class PricesSpider(scrapy.Spider):
    name = "prices"

    def start_requests(self):
        yield scrapy.Request("https://t7proxy.com/ip", meta={"proxy": "http://LOGIN:PASSWORD@HOST:PORT"})

    def parse(self, response):
        yield {"ip": response.text.strip()}

# pip install playwright && playwright install chromium
from playwright.sync_api import sync_playwright

with sync_playwright() as p:
    browser = p.chromium.launch(proxy={
        "server": "http://HOST:PORT",  # the HTTP port
        "username": "LOGIN",
        "password": "PASSWORD",
    })
    page = browser.new_page()
    page.goto("https://t7proxy.com/ip")
    print(page.inner_text("body"))
    browser.close()

SOCKS5 and HTTP(S) on each plan →

Collect public data responsibly

Proxies are for lawful data work. A few rules keep your crawls fair to the sites you visit:

• Collect public pages only. Do not sign in to accounts or go past login pages that are not yours.
• Read the site's terms and robots.txt, and follow them.
• Keep request rates reasonable: add delays, limit parallel requests per site, and slow down when a site answers slowly or returns errors.
• Leave personal data out of what you collect.
• Respect copyright in what you store and publish.

Read our acceptable use rules →

Proxies for AI data collection, bought from an AI assistant

Teams that build datasets for research and AI use the same setup: public pages, a polite crawl rate, and proxies in the countries the data has to represent. Datacenter IPv4 keeps large jobs at a fixed cost per IP; residential adds region and city targeting.

You can also buy and manage proxies from Claude, ChatGPT, Cursor or VS Code. Connect the T7Proxy MCP server, ask for a quote, and the assistant confirms the total with you before it buys from your wallet.

Connect the T7Proxy MCP server →

Check your proxies before a crawl

Before a long run, test your list. The free proxy checker takes up to 10 proxies at a time and shows whether each one works, with its exit IP, country, ISP and latency.

Test your proxy list for free →SEO proxies for rank tracking →