GitHub Overhauls Status Page With New Degraded Performance Tier

Peter Zhang Apr 17, 2026 16:41

GitHub introduces three-tier incident classification, per-service uptime metrics, and dedicated Copilot AI model tracking in major status page transparency update.

GitHub Overhauls Status Page With New Degraded Performance Tier

GitHub rolled out significant changes to its status page on April 17, 2026, introducing a new incident classification system and per-service uptime metrics that give developers more granular visibility into platform health.

The update comes after the company acknowledged availability issues earlier this year, and represents a direct response to developer complaints about unclear incident reporting.

Three-Tier Incident Classification

The most notable change is a new "Degraded Performance" severity level sitting below Partial Outage and Major Outage. Previously, even minor service hiccups triggered a Partial Outage classification—misleading users into thinking services were down when they were merely slow.

Under the new system, Degraded Performance covers situations where services remain operational but impaired—think elevated latency, reduced functionality, or intermittent errors hitting a small percentage of requests. Partial Outage now indicates significant unavailability for a meaningful number of users, while Major Outage signals broad unavailability affecting most or all users.

Uptime Metrics Get Specific

GitHub now publishes 90-day uptime percentages for each individual service on its status page. The calculation weights incidents by severity: Major Outages count at 100% downtime, Partial Outages at 30%, and Degraded Performance at 0%.

A practical example: a 1-hour Partial Outage translates to 18 minutes of effective downtime in the calculation, not the full hour. Degraded Performance incidents don't impact uptime percentages at all—the service technically stayed functional.

Copilot Gets Its Own Incident Tracking

A new "Copilot AI Model Providers" component now separates model-specific issues from broader Copilot outages. Previously, when a single AI model provider went down, GitHub declared an incident against all of Copilot—even when features like Chat and cloud agent could fall back to alternative models.

For teams using auto model selection, this distinction matters. A single model going offline doesn't necessarily mean Copilot stops working.

Why This Matters for Dev Teams

For engineering organizations running CI/CD pipelines through GitHub Actions or depending on Copilot for daily coding workflows, clearer incident classification helps with capacity planning and incident response decisions. Knowing whether GitHub is experiencing degraded performance versus an actual outage affects whether you wait it out or switch to backup processes.

The 90-day uptime metrics also give procurement and security teams hard numbers for vendor assessments—something GitHub previously kept closer to the vest.

Image source: Shutterstock