Grok outage, 6 September 2026
Grok was down for 20 minutes on 6 September 2026, from 04:45 to 05:05 UTC, as measured by HostTracker's checkpoints.
What our checkpoints saw
ALL TIMES UTCThe first failed check is the top row, and the recovery is the last one. A row marked "x N" stands for several consecutive checks with the same result from that checkpoint.
| Time (UTC) | Location | Result | Error |
|---|---|---|---|
| 6 Sep 04:55 | Saint Petersburg, Russian Federation | Failed x 2 | Http 403 |
The shaded band is the outage window. Points are averaged per bucket across the checkpoints that answered, so a gap in the line is a period where nothing answered at all.
This is the most recent outage we recorded.
Does your product depend on Grok?
An outage like this one reaches your users through your product, and most teams hear about it from a customer rather than from a dashboard. HostTracker watches any endpoint you name - your own site, an API you call, a vendor you resell - from 300+ locations, confirms a failure from several of them before it says anything, and then tells you by email, SMS, voice call or your chat tool.
You get the same record you are reading here for your own services: every outage timestamped, with the checks that observed it.
Independent measurement by HostTracker. Not affiliated with Grok.