Rotating 4G/5G mobile proxies for scraping hard targets

Why do sites block mobile IPs less often?

Because one mobile IP usually stands for many real people. Carriers run carrier-grade NAT: thousands of phones share a small set of public addresses, and RFC 6598 even reserves the range 100.64.0.0/10 for the inside of these networks. A site that blocks one mobile IP can lock out every subscriber behind it, so most sites block mobile ranges with care.

The same sharing has a cost for you. A mobile IP can arrive with a history created by other users of the carrier, and its speed depends on signal and cell load. That is why mobile traffic is a tool for specific targets, not a default.

Is mobile worth the higher price per GB?

Only where residential fails. CrawlRelay residential traffic costs $0.50/GB from 50 GB; mobile costs $1.50/GB. In Proxyway's Proxy Market Research 2026, the median success rate on real targets was 75.26% for mobile proxies and 74.43% for residential ones: across all providers, mobile was not much better on average.

The gain is target-specific. Measure it before you pay for it:

1. Run 200 to 500 URLs of one target through residential and through mobile.
2. Count usable responses, not HTTP 200s: a block page can return 200.
3. Divide each run's traffic cost by its usable responses.
4. Route a domain to mobile only if its cost per usable response is lower there.

How do I route only some domains to mobile IPs?

Keep two sets of credentials, one per plan, and choose per request. The rest of your scraper stays the same: same gateway, same username syntax, same retry logic.

from urllib.parse import urlsplit

import requests

GATEWAY = "gw.crawlrelay.com:7777"
RESIDENTIAL = ("RES_USERNAME-country-us", "RES_PASSWORD")
MOBILE = ("MOB_USERNAME-country-us", "MOB_PASSWORD")
MOBILE_HOSTS = {"m.example.com", "app.example.org"}  # hosts where residential measured worse


def proxies_for(url):
    user, password = MOBILE if urlsplit(url).hostname in MOBILE_HOSTS else RESIDENTIAL
    proxy = "http://%s:%s@%s" % (user, password, GATEWAY)
    return {"http": proxy, "https": proxy}


for url in ["https://www.example.com/item/1", "https://m.example.com/item/1"]:
    r = requests.get(url, proxies=proxies_for(url), timeout=(10, 45))
    print(url, r.status_code, len(r.content))

How does rotation work on the mobile pool?

The same way as residential: a username without a session ID gets a new carrier IP on each new connection, and -session-<id>-ttl-<minutes> keeps one IP across connections. Two differences matter in practice:

• Carriers change IPs on their own. A sticky mobile session can lose its IP earlier than its TTL when the device reconnects to the network.
• Latency varies more. Give mobile requests a longer read timeout, about 45 to 60 seconds, and retry on a new connection instead of waiting longer.