VPNLab buys accounts, tests 30 days, and publishes comparable measurements. Full protocol lives on Methodology; this guide expands the consumer-facing version.
We measure download/upload on nearby and distant servers using consistent baseline connections. WireGuard/NordLynx-class protocols usually win latency.
DNS, IPv6, and WebRTC leak checks matter; a “connected” UI means little if DNS leaves the tunnel.
We verify major platforms and read jurisdiction + audit claims critically. Marketing “no logs” language is not the same as audited no-logs.