GitHub's status page logged an incident affecting Grok 4.6 and Grok 4.7 service in Copilot starting at 23:41:54 UTC on October 23, 2026, and marked it resolved at 00:23:32 UTC on October 24, a gap of 42 minutes. The listing gives no description of symptoms, no error codes, and no indication of how many requests or users were touched. For on-call staff who rely on Copilot's Grok-backed features during the overnight window, that stretch of time is the entire public record so far.
GitHub has not disclosed what caused the disruption, which Copilot features depended on Grok 4.6 or 4.7 during the incident, or whether the fix involved a rollback, a configuration change, or a capacity adjustment on the vendor side. The status page format typically marks incidents resolved once monitoring shows recovery, not once a root cause is confirmed, so the 42-minute window reflects restoration of service rather than an explanation of the failure. GitHub did not disclose a root cause analysis timeline either, and no follow-up post has been referenced alongside the resolved entry.
Outages tied to specific model versions inside a shared platform like Copilot tend to be narrow by design, since providers often route requests to particular models or model generations rather than a single shared backend. That architecture means an incident naming Grok 4.6 and 4.7 specifically is plausibly isolated from other Copilot functionality, such as completions powered by different models, though GitHub's status entry does not state this explicitly. Readers should treat that separation as a reasonable inference from how multi-model platforms are typically built, not as a confirmed fact from the incident record.
Teams running on-call rotations against Copilot integrations should check whether any automation, CI steps, or internal tooling pinned to Grok 4.6 or 4.7 logged failures or timeouts between 23:41 and 00:23 UTC on October 23 into October 24. Logging systems that capture model identifiers in request metadata will make this correlation straightforward; those that do not will need to rely on timestamp overlap alone. Anyone who paged during that window, or who saw Copilot latency or errors without an obvious internal cause, now has a plausible external explanation on record, even without a root cause.
What remains open is whether the incident recurred, whether it affected only a subset of regions or accounts, and whether GitHub plans to publish more detail beyond the status page entry. None of that has been addressed publicly as of this writing. The resolved marker closes the clock on the outage itself, but it leaves the underlying question, what actually failed inside Grok-backed Copilot service for those 42 minutes, unanswered for now.