Oxylabs
Shifter's IPs came from up to 59% more separate networks, and more of them were online in 2 of 5 countries.
Every provider publishes a pool size. Almost none publish how many of those addresses are reachable in the country you need, at the moment you need them. So we measure it ourselves, the same way for every network, and publish the method alongside the results.
About these numbers. Every network here was tested in the same five countries only: the United States, United Kingdom, France, Canada and Germany. The figures are the distinct IPs each network had live and answering during the test window, not the size of its network. Residential IPs rotate all day, so every provider serves far more addresses across 24 hours than any single window can capture.
Shifter's IPs came from up to 59% more separate networks, and more of them were online in 2 of 5 countries.
Similar size pool, wider spread: up to 66% more networks, and more IPs online in 1 of 5 countries.
More working IPs than SOAX everywhere we looked, peaking at 477% more (Germany).
Shifter had more IPs online in all 5 countries, up to 282% more (France).
Similar size pool, wider spread: up to 54% more networks, and more IPs online in 1 of 5 countries.
More working IPs than Massive everywhere we looked, peaking at 1400% more (France).
Shifter's IPs came from up to 236% more separate networks, and more of them were online in 4 of 5 countries.
Similar size pool, wider spread: up to 208% more networks, and more IPs online in 4 of 5 countries.
More working IPs than NodeMaven everywhere we looked, peaking at 305% more (France).
Shifter had more IPs online in all 5 countries, up to 440% more (United States).
Ahead in every country we tested, with the widest gap at 183% more IPs online (Germany).
More working IPs than ProxyEmpire everywhere we looked, peaking at 1321% more (France).
Shifter had more IPs online in all 5 countries, up to 3478% more (Germany).
Ahead in every country we tested, with the widest gap at 66% more IPs online (France).
Each benchmark counts the IPs a network had live and reachable during the test window, not the size of its network. Residential IPs rotate constantly as real devices come online and drop off, so every provider here serves far more addresses across a day than any single test can see. What limits a crawl is how many respond at the same moment, and that is what we measure. Every provider gets identical request counts, concurrency and targets, tested one at a time from the same machine.
Our numbers are a starting point, not a verdict. Measure both networks on the sites you actually scrape, in the countries you actually need.