Operational · last checked 04:14 UTC · 941 ms · checked every about 5 minutes
Uptime
24 hours
72.17%
7 days
67.60%
30 days
80.41%
90 days
81.83%
90 days agotoday
Handshake latency p50 2747 ms · p95 4948 ms · mean time to recover 364 min
Incidents (90 days)
Outage · 2026-10-20 19:59 UTC · 6.0 h · rate_limit
Outage · 2026-10-20 05:46 UTC · 3 min · http_5xx
Outage · 2026-10-19 10:54 UTC · 18.0 h · rate_limit
Outage · 2026-10-18 18:09 UTC · 6.0 h · rate_limit
Outage · 2026-10-17 20:20 UTC · 6.0 h · rate_limit
Outage · 2026-10-16 20:54 UTC · 6.1 h · rate_limit
Outage · 2026-10-15 19:16 UTC · 6.1 h · rate_limit
Outage · 2026-10-14 17:28 UTC · 6.1 h · rate_limit
Outage · 2026-10-13 21:22 UTC · 6.1 h · rate_limit
Outage · 2026-10-11 20:33 UTC · 6.1 h · rate_limit
Outage · 2026-10-09 19:54 UTC · 6.1 h · rate_limit
Outage · 2026-10-08 03:59 UTC · 7 min · timeout
Outage · 2026-10-07 19:36 UTC · 6.1 h · rate_limit
Outage · 2026-10-07 02:33 UTC · 14 min · timeout
Outage · 2026-10-06 18:53 UTC · 6.1 h · rate_limit
Outage · 2026-10-05 18:18 UTC · 6.1 h · rate_limit
Outage · 2026-10-04 14:07 UTC · 8 min · timeout
Outage · 2026-10-04 12:45 UTC · 16 min · timeout
Outage · 2026-10-02 20:43 UTC · 6.1 h · rate_limit
Outage · 2026-10-01 19:34 UTC · 6.1 h · rate_limit
Outage · 2026-10-01 15:33 UTC · 7 min · timeout
Outage · 2026-09-30 19:32 UTC · 6.1 h · rate_limit
Outage · 2026-09-30 17:10 UTC · 7 min · http_5xx
Outage · 2026-09-29 19:22 UTC · 6.1 h · rate_limit
Outage · 2026-09-27 23:31 UTC · 6.2 h · rate_limit
Outage · 2026-09-25 22:24 UTC · 6.2 h · rate_limit
Outage · 2026-09-22 21:22 UTC · 6.2 h · rate_limit
Outage · 2026-09-21 20:23 UTC · 6.1 h · rate_limit
Outage · 2026-09-19 23:01 UTC · 6.1 h · rate_limit
Outage · 2026-09-17 18:05 UTC · 6.2 h · rate_limit
How this is measured
Checked by Glimind's probe (initialize + tools/list, never a tool call) from one region. Uptime counts a server as up when it completes the handshake, including when it asks for credentials. An outage starts at the first failed check, so short blips between checks are not seen. Every server is measured the same way; an owner's plan never changes these numbers. Owner updates are the owner's own words and sit next to the measurements, not in place of them.