GitHub January 2026: Two Incidents, Copilot Outage

TL;DR

• GitHub Copilot went down for 46 minutes on Jan 13 due to a config error during a model update; error rates hit 100%
• Jan 15 saw 1 hour 40 minutes of degraded performance across issues, PRs, API, and login due to a data store infrastructure upgrade
• Both incidents are now under root-cause review with new safeguards and monitoring in place

What Dropped

GitHub published its January 2026 availability report, documenting two separate incidents that impacted core services. Copilot Chat experienced a complete outage on January 13, while a broader infrastructure issue on January 15 affected pull requests, issues, API, and authentication across the platform.

The Dev Angle

January 13 incident: GitHub Copilot went offline from 09:25 to 10:11 UTC (46 minutes) with error rates averaging 18% and spiking to 100%. The culprit was a configuration error introduced during a model update. The initial rollback mitigated the issue, but recovery extended another 35 minutes because OpenAI's GPT-4.1 model was also experiencing degraded availability upstream. This affected Copilot Chat across VS Code, JetBrains IDEs, and other integrations.

January 15 incident: A more widespread outage lasted 1 hour 40 minutes (16:40–18:20 UTC) and hit issues, pull requests, notifications, Actions, repositories, the API, and account login. An average of 1.8% of combined web and API requests failed, peaking at 10%. The root cause was an infrastructure upgrade to a major version of their data stores that created unexpected resource contention under load. Unauthenticated users were hit hardest, but authenticated users also experienced timeouts and latency.

GitHub's response: stronger monitoring, improved test environments for configuration changes, and better validation processes for infrastructure upgrades to catch issues under high load before full release.

Should You Care?

If you rely on Copilot for daily development, the January 13 outage is a reminder that AI-assisted coding tools have their own infrastructure dependencies—and when they fail, they fail hard. The 46-minute window is significant for teams using Copilot as a core productivity tool.

The January 15 incident is more relevant to everyone. API failures, login issues, and PR/issue degradation affect the entire GitHub workflow. A 1.8% average failure rate might sound small, but it compounds across millions of requests. If you were pushing code or reviewing PRs during that window, you felt it.

The good news: both incidents are now documented with clear root causes and mitigation plans. GitHub is investing in better pre-release validation and faster detection. If you're building on GitHub's API or relying on Copilot in production workflows, monitor the GitHub Status page for real-time updates and post-incident recaps.

For context on how AI tooling like Copilot is reshaping developer workflows, see how AI is reshaping developer choice in the latest Octoverse data.

Source: GitHub Blog