How fast is the world's internet?

Independent servers, not your provider's own. Many speed tests measure against a server sitting inside your internet provider's own network, the shortest and least congested path, which flatters the result. SpeedOf.Me tests against independent, third-party CDN edge servers, the same infrastructure that delivers the websites, video and downloads you use every day, so the number reflects real browsing performance rather than a best-case local loop. Providers have offered to host our test servers free of charge for over a decade; we have always declined and paid for neutral capacity instead, an approach we helped pioneer when SpeedOf.Me launched in 2011.

One stream, the way real downloads behave. We use a single connection that starts with small files and ramps upward, reporting the sustained speed of the largest sample your line can pull. Tests that open many parallel streams report a higher, more theoretical peak a line can reach in ideal conditions.

The same ruler, everywhere, since 2011. We have kept this method as unchanged as possible for over a decade. That consistency is the point: our numbers stay comparable across countries and across years because every test, everywhere, is measured the same way. Every figure here is produced by the SpeedOf.Me measurement engine, the same engine developers embed through our speed test API.

Why the order can differ, not just the level. An on-net, multi-stream test rewards the size of the local pipe; an off-net, single-stream test rewards the whole route to the content people actually use. A country can be world-class on the first ruler and mid-table on the second: Singapore tops Ookla's fixed-broadband index, yet a single stream across the open internet lands it far lower here. Neither ranking is wrong; they answer different questions. The full comparison, ruler by ruler: What is Real-World Speed?