Return to Live Telemetry

Incident Archive

A permanent, immutable record of system performance. We believe that hiding failures prevents trust.

Date (UTC)
Status
Incident Details
Duration
Nov 14, 202508:42 UTC
OK
System Wide

All Systems Operational

Daily automated integrity check completed. No anomalies detected in inference clusters or safety layers.

ID: INC-2025-114
N/A
Oct 30, 202514:20 UTC
Degraded
API Gateway

Elevated Latency in EU-West

We observed a 45ms latency increase for traffic routing through Frankfurt. Traffic was automatically re-routed to Paris edge nodes within 12 seconds.

ID: INC-2025-098
12m
Oct 15, 202503:00 UTC
Maint
Core Cluster

Scheduled Maintenance: Vector DB Upgrade

Planned upgrade of the long-term memory retrieval stack. Zero downtime observed due to blue/green deployment strategy.

ID: INC-2025-082
45m
Sep 22, 202519:45 UTC
Outage
Safety Filter

False Positive Spike in Constitutional Layer

An update to the "Self-Harm" refusal circuit caused a 2% spike in false refusals for medical queries. Model weights were rolled back to v2.4.1 immediately.

ID: INC-2025-071
18m
Aug 10, 202511:15 UTC
Degraded
Dashboard

Billing Metrics Delay

Usage logs for the "Reasoning-Alpha" endpoint were delayed by 15 minutes in the dashboard. No data was lost; purely a display latency.

ID: INC-2025-055
2h
Jul 04, 202509:00 UTC
Maint
Infrastructure

H100 Cluster Expansion

Added 4,096 H100 GPUs to the primary training cluster. Compute capacity increased by 40%.

ID: INC-2025-012
4h