Which VPN is fastest?

None is fastest for everyone. A VPN's speed depends on how far you are from its server, on your own internet line, on how busy the server is and on the time of day. The VPN that is fastest for one person can be slow for another. This page explains what decides it, how we measure speed, and where to compare our measurements yourself, one test machine at a time.

When a list names one VPN the fastest, it is really telling you which one was fastest on one line, to one set of servers, at the times its tests ran. Our tests can't tell you more than that either. That is why we show each test machine's results on their own and let you do the sorting.

What decides a VPN's speed

Distance to the server. Everything you send goes to the VPN server and back. The farther away the server is, the longer each trip takes, and the slower a download gets up to speed. A server near you usually beats a faraway one, whoever runs it.
Your own line. A VPN can't be faster than the connection under it. On a 50 Mbps line, two VPNs that could each carry 300 Mbps will feel the same.
How busy the server is. You share a VPN server with the provider's other customers. A crowded server is a slower one.
The time of day. Evenings are busy, at home and on the servers. The same server can be slower at 9 pm than at 9 am.

Every tunnel we test uses WireGuard, so the protocol is the same in all our tables.

How we measure speed

Each of our nodes is a machine of ours on one internet line. About once a day it runs a speed test through every tunnel it holds: one download of 50 MB and one upload of 15 MB, one connection each way, from Cloudflare's speed-test server near where the tunnel comes out. Just before each test, the node runs the same test on its own line, outside any tunnel. That gives every result a baseline: what the line could do at that moment. The tables show the middle value of the tests over the last 7 days, about 7 per server. Each figure lists how many tests are behind it in its row's details.

What this shows: how tunnels compare when they are tested the same way, from the same line. What it doesn't show: the most a connection can carry. Over a long distance, a single transfer of that size is still speeding up when it ends, so it reads lower than the line could do. Compare rows within one table, not with a speed test in your browser. The full details are in Methodology: the speed test, the baseline, and what a speed test does and does not show.

Compare for yourself

Each node has its own table. A figure depends on that node's line and on how far the node is from each server, so compare rows within one table only. Pick the node closest to your own situation. Each link opens a table already sorted. Sorting puts one column in order and nothing else: it is not a ranking. Latency is the middle of the last 24 hours of checks, taken every 5 minutes, not part of the speed test.

Asia Residential, home line, 1 Gbps Down / 500 Mbps Up: download, fastest first · upload, fastest first · latency, lowest first
Asia Data Center, data center line, 4 Gbps Out / 40 Gbps In: download, fastest first · upload, fastest first · latency, lowest first
South Africa Residential, home line, 200 Mbps Down / 100 Mbps Up: download, fastest first · upload, fastest first · latency, lowest first

Each row's details show the node's own line, measured just before the tests, beside the tunnel's figures, as two plain numbers.

To measure your own speed with and without a VPN, see how to test your VPN's speed.

What our data shows

Every node's table is on the results page, with what each column means. Each number has the checks behind it, one click away in its row. See also how we measure and its known limits.

Coming soon

Our own speed test, to measure your own connection with the VPN off and on.