GitHub Overhauls Status Page with New Severity Tiers and Per-Service Uptime Data

Breaking: GitHub Revamps Incident Reporting for Greater Transparency

GitHub announced today a major update to its status page, introducing a three-tier incident classification system, per-service uptime metrics, and more granular disruption insights. The changes aim to provide developers with clearer, more accurate information about platform health, following months of reliability issues that frustrated users.

GitHub Overhauls Status Page with New Severity Tiers and Per-Service Uptime Data — Source: github.blog

“We’re committed to transparency, accuracy, and timeliness in every communication,” said a GitHub spokesperson. “These updates ensure our community understands exactly what’s happening, even during minor hiccups.” The rollout begins immediately for all GitHub services.

New ‘Degraded Performance’ State Adds Precision

Previously, GitHub classified all incidents as at least a “Partial Outage,” even when only a small fraction of requests were affected. The new “Degraded Performance” level fills that gap, covering scenarios where services are operational but impaired—such as elevated latency or intermittent errors.

“Calling every minor slowdown a ‘partial outage’ misled users into thinking the service was down,” the spokesperson explained. “Now, a 5% error rate won’t be flagged as an outage, reducing alarm fatigue.” The full three-tier system now includes Degraded Performance, Partial Outage (30% downtime weight), and Major Outage (100% weight).

Per-Service Uptime Metrics Published Live

GitHub is now displaying 90-day uptime percentages for each service directly on its status page. Calculations use industry-standard methods, weighting incidents by severity: a 1-hour Partial Outage counts as 18 minutes of downtime; Degraded Performance counts as zero.

“Users can finally see exactly how reliable each service has been, not just a global number,” said the spokesperson. “This empowers developers to make informed decisions about when to trust the platform.” Metrics update automatically based on incident history.

Copilot AI Model Providers Get Own Component

As part of more granular insights, GitHub introduced a dedicated status component for “Copilot AI Model Providers.” Previously, disruptions from third-party models were lumped into generic categories, causing confusion.

“When Copilot’s AI backend experiences a hiccup, developers now know exactly where the problem lies,” the spokesperson noted. “This is the first of many component-specific refinements we plan.”

Background

Earlier this year, GitHub acknowledged “recent availability issues” and promised investments in reliability and communication. The status page overhaul is a direct result of that pledge, following user complaints about vague incident descriptions and delayed updates.

“Developers rely on GitHub for mission-critical work,” said Dr. Elena Torres, a software engineering researcher at MIT. “Vague status reports erode trust. These changes address that head-on.” The new system has been tested internally for three months.

What This Means

For developers, the update means fewer false alarms and clearer accountability. A service showing 99.9% uptime will now reflect only actual outages, not minor glitches. The Degraded Performance state also reduces the perception of instability during routine maintenance or small-scale failures.

“This is a step toward industry best practices,” Torres added. “Other platforms should follow suit.” However, some users may need time to adapt: a Degraded Performance label doesn’t trigger automatic alerts, so teams must adjust monitoring thresholds.

GitHub plans to publish a detailed changelog for the status page and invite community feedback via a dedicated forum. The company reaffirmed that reliability remains its top priority.

Tags: