PINGRANK.GG

LEADERBOARD

Networks ranked by real in-game latency. Every row is an (ISP, game, access class) aggregate built from player-submitted sessions; pick a game to rank networks head-to-head.

all games
# network regionwhere this network's players connect from — the country the ISP's addresses are registered in. Derived on the server from the submitting connection; the address itself is never stored. more → game access?how your ISP hands you the Internet: a public address of your own, or a shared/translated one (CGNAT, NAT64/464XLAT, DS-Lite). Detected by the CLI from reflector evidence; rankings split by it because address sharing changes how connections behave. full explainer → p50the typical ping: half of all measured pings were faster than this, half slower. This is what the game feels like most of the time — lower is better. more → p95the rough-moment ping: 19 of every 20 pings were faster than this. When p95 sits far above p50, the connection is fine most of the time but has lag spikes. more → jitterhow much the ping wobbles from moment to moment. A steady connection feels smoother than a fast but jumpy one — high jitter is what makes a game feel stuttery. more → lossthe share of data packets that never arrived and had to be resent or skipped. Even 1–2% can cause rubber-banding and shots that don't register. more → relaythe share of traffic that travelled through the game's own relay servers (a middleman, like Valve's network) instead of straight to the game server. Counted separately — never mixed into the ping numbers. more → sessionshow many recorded play sessions this row is built from. More sessions means more players backing the numbers — and more trustworthy rankings. more →
1 SPINTEL-AS-AP Spintel Pty Ltd — Rocket League unknown 12.0 ms 12.3 ms 1.4 ms 0.8% 0% 1
Headline latency blends clean measurement methods only (direct, protocol, tcp-estats); last-hop and mixed segments are counted but never averaged in. Relay% is the share of segments measured to a relay edge (Valve SDR and similar), reported separately — never passed off as server latency.