Resolved -
We've had several spikes of incidents and we have mitigated them. The last two ones being:
* between 5:47 PT / 12:47 UTC and 6:23 PT / 13:23 UTC
* a period of medium level of errors
* between 7:28 PT / 14:28 UTC and 8:35 PT / 15:35 UTC
The issue has been fully mitigated.
Jul 25, 16:37 UTC
Identified -
We're seeing another spike of errors as of 5:48 PT / 12:48 UTC.
Jul 25, 12:54 UTC
Monitoring -
The last bump of errors was between 1:58 PT / 8:58 UTC and 2:21 PT / 9:21 UTC.
We haven't yet resolved the root cause, and we are working on that. We are applying tactical mitigations and we're still at a high risk of errors.
Jul 25, 09:26 UTC
Identified -
We're seeing elevated errors again since 1:58 PT / 8:58 UTC.
Jul 25, 09:03 UTC
Monitoring -
The level of errors has been reduced to a low rate since 1:00 PT / 8:00 UTC. However we still are at risk of the errors coming back.
Jul 25, 08:07 UTC
Investigating -
We are currently investigating an issue on Claude 4 Sonnet that started at 23:55 PT / 6:55 UTC.
Jul 25, 07:11 UTC