Mirror comparison matrix
Side-by-side comparison of the six mirrors. Score is uptime + 100/(p50 latency in ms), ranked top first. Pick the top row for a low-friction session.
torzon4v7bcakvo7qikd…
tv4pfwlnoezgtzwrr33f…
otw35cxf2rssl23tsvtq…
ncmebasolcj2pmw5oy2b…
evwigej45n3nywbn3aqd…
dgkozv5myc3lfpedjl2k…
Why this scoring
The formula prioritises uptime first (a mirror that is down 5% of the time is useless), then latency (a mirror that responds in 100ms feels twice as fast as one that responds in 200ms). p95 breaks ties when uptime is close.
Ranking is not stable across the day
Tor circuits vary. A mirror ranked #1 at noon may drop to #3 at midnight because the guard node happens to be on a different route. The comparison here uses 30-day aggregates, which smooths out short bumps. For an in-the-moment pick, try the top two and see which opens first.
Age is a secondary signal
Older mirrors have accumulated known-good Tor descriptors and tend to route via reliable guards. Fresh mirrors sometimes bounce for the first week after introduction. If two mirrors score close, prefer the older one.