Proxies for web scraping, paid in crypto with no KYC

Residential or mobile proxies for web scraping?

Use residential proxies for almost all scraping. They cost less per GB, rotate per connection and cover country and city targeting. Keep 4G/5G mobile proxies for the few mobile-first targets that challenge home IPs too.

No proxy can promise access. Proxyway's 2026 research measured a 99.28% median infrastructure success rate for residential pools, but 74.43% on real targets such as Amazon, Google and Instagram. Your request rate, headers and retry logic matter as much as the IP.

Rotating or sticky sessions for scraping?

Rotate for independent pages, stick for flows. A crawler fetching thousands of unrelated product pages should get a new exit per connection: no session ID in the username. A flow that must look like one visitor, such as paginated search results or a cart, needs a sticky session: add -session-{id}-ttl-{minutes} and reuse the ID.

Remember that rotation is per connection. A requests.Session keeps its HTTPS connection alive, so it stays on one exit until that connection closes.

How many requests can I run in parallel?

Start low and raise concurrency per target while the error rate stays flat. The right number depends on the target site far more than on the proxy. A practical setup:

• Concurrency: begin with a few parallel connections per domain, then increase step by step.
• Retries: retry on 403, 429 and 5xx with exponential backoff; a retry on a new connection also gets a new exit IP.
• Pacing: honour Retry-After headers and crawl-delay hints; a slower crawl that completes beats a fast one that gets blocked.
• Sub-users: give each scraper its own credentials and traffic cap, so one runaway job cannot drain the balance.

How do I scrape with Python requests?

The function below fetches a URL through a rotating Kenyan exit, retries with exponential backoff, and identifies the crawler honestly in its User-Agent. More patterns (httpx, aiohttp, SOCKS5) are in the Python proxy guide.

import random
import time
import requests

GATEWAY = "gw.namelessproxy.com:7777"
USERNAME, PASSWORD = "USERNAME", "PASSWORD"
RETRY_CODES = {403, 429, 500, 502, 503, 504}
HEADERS = {"User-Agent": "PriceResearchBot/1.0 (+https://example.com/bot)"}

def fetch(url, country="ke", tries=4):
    proxy = f"http://{USERNAME}-country-{country}:{PASSWORD}@{GATEWAY}"
    for attempt in range(tries):
        try:
            r = requests.get(url, proxies={"http": proxy, "https": proxy},
                             headers=HEADERS, timeout=30)
            if r.status_code not in RETRY_CODES:
                return r
            ra = r.headers.get("Retry-After", "")
            wait = int(ra) if ra.isdigit() else 0
        except requests.RequestException:
            wait = 0
        # New connection on retry = new exit IP
        time.sleep(max(wait, 2 ** attempt + random.random()))
    raise RuntimeError(f"giving up on {url}")

print(fetch("https://example.com/").status_code)

How do I set up Scrapy with rotating proxies?

Add a small downloader middleware that writes the proxy URL into request.meta; Scrapy's built-in proxy middleware then sends the credentials. Requests that share a session value in their meta keep the same exit IP.

# middlewares.py
GATEWAY = "gw.namelessproxy.com:7777"
USERNAME, PASSWORD = "USERNAME", "PASSWORD"

class NamelessProxyMiddleware:
    def process_request(self, request, spider):
        user = f"{USERNAME}-country-{request.meta.get('country', 'eg')}"
        session = request.meta.get("session")  # same value = same exit IP
        if session:
            user += f"-session-{session}-ttl-10"
        request.meta["proxy"] = f"http://{user}:{PASSWORD}@{GATEWAY}"

# settings.py
ROBOTSTXT_OBEY = True
CONCURRENT_REQUESTS_PER_DOMAIN = 4
AUTOTHROTTLE_ENABLED = True
RETRY_TIMES = 3
RETRY_HTTP_CODES = [403, 429, 500, 502, 503, 504]
DOWNLOADER_MIDDLEWARES = {
    "myproject.middlewares.NamelessProxyMiddleware": 350,  # runs before HttpProxyMiddleware (750)
}

How much traffic does scraping use?

Measure your average response size, then divide. Traffic is billed in decimal GB, so if an average page weighs 500 KB, 1 GB covers about 2,000 pages; at 100 KB it covers about 10,000. Requesting JSON endpoints instead of full pages, and blocking images and fonts in headless browsers, cuts usage further.

Test a target with the 1 GB pack first, then buy a larger pack: prices fall to $0.53/GB on the 1 TB pack, and unused traffic never expires. All tiers are on the pricing page.

Is web scraping legal?

It depends on the data, the method and the country; this is not legal advice. US courts have reached different results depending on the claim (contract, computer-access law or copyright/DMCA):

• hiQ v. LinkedIn: in April 2022 the Ninth Circuit found that scraping public data was unlikely to be access "without authorization" under the CFAA. In November 2022 a district court still found that hiQ had breached LinkedIn's User Agreement.
• Meta v. Bright Data: in January 2024 a federal judge ruled that Meta's terms did not bar logged-off scraping of public data. The ruling covered the contract claim only.
• Google v. SerpApi: Google alleged DMCA circumvention in December 2025. On July 20, 2026, the court dismissed the DMCA claims over ordinary search results, with leave to amend only on one narrower theory.
• Reddit v. SerpApi and Perplexity: on July 31, 2026, most claims survived a motion to dismiss. That is a pleading-stage ruling, not a finding of liability.

Personal data needs extra care. Under the GDPR, personal data includes names, location data and online identifiers, even on public pages, and must be limited to what is necessary for the purpose.