AI Incidents
Back to incident register
Source reviewedOperational incidentMedium

xAI reported a multi-platform Grok model outage lasting more than three hours

xAI's official status pages documented a model outage beginning at 13:30 UTC on September 3 across its API and Grok product surfaces. The us-east-1 API status record marked traffic healthy again after three hours and thirty-seven minutes.

First observed
Sep 3, 2026, 3:30:00 PM
Disclosed
Sep 3, 2026, 3:30:00 PM
Status
Resolved
Confidence
98%
Organization
xAI
Last reviewed
Sep 4, 2026

Observed

Facts supported by sources

  • xAI reported a Grok model outage affecting the us-east-1 API and multiple Grok application surfaces beginning at 13:30 UTC on September 3.
  • The us-east-1 API incident was marked resolved at 17:07 UTC after three hours and thirty-seven minutes, with xAI stating that traffic was healthy again.

Assessment

Interpretation, not observation

  • xAI did not disclose a technical root cause or affected-user count in the status records reviewed.
  • The outage overlapped with incidents at other AI providers, but no shared cause is established by the official sources.

Impact

Impact

Unavailable or failing Grok model requests across xAI API and consumer product surfaces during the outage window.

Response

Response

xAI investigated the outage, restored traffic and marked the affected API and product surfaces healthy again.

Methodology: Facts, interpretations, and uncertainty are kept separate. Confidence describes the strength of the evidence, not a probability estimate.