GitHub Copilot Hit 100% Error Rate During January Outage Tied to OpenAI GPT-4.1

GitHub’s January 2026 availability report, revealed February 11, particulars two important incidents that disrupted providers for builders worldwide—together with a Copilot outage the place error charges briefly hit 100% as a result of problems with OpenAI’s GPT-4.1 mannequin.

The primary incident struck on January 13 between 09:25 and 10:11 UTC. GitHub Copilot’s error charges averaged 18% in the course of the 46-minute window, spiking to finish failure at factors. Chat options throughout VS Code, JetBrains IDEs, and different built-in merchandise went down.

A configuration error throughout a mannequin replace triggered the preliminary downside. GitHub rolled again the change, however restoration stretched till 10:46 UTC as a result of OpenAI was concurrently experiencing degraded availability for GPT-4.1. The dependency chain failure highlights ongoing reliability issues as AI coding assistants change into embedded in developer workflows.

Two days later, on January 15, GitHub confronted broader infrastructure troubles. Between 16:40 and 18:20 UTC, customers encountered elevated latency and timeouts throughout points, pull requests, notifications, Actions, repositories, API entry, and account login. Mixed internet and API request failures averaged 1.8%, briefly peaking at 10% early within the incident.

The wrongdoer? An infrastructure replace to GitHub’s knowledge shops. Upgrading to a brand new main model created sudden useful resource rivalry beneath manufacturing load—the type of downside that does not floor in testing environments. Unauthenticated customers bore the brunt of the influence, although authenticated classes weren’t spared. Rolling again to the earlier secure model resolved the difficulty.

January proved tough for GitHub past these two headline incidents. Authentication providers hiccuped on January 22 for practically an hour. Repository creation and cloning degraded between January 24-25 as a result of database latency. Actions workflows ran a mean 49 seconds late on January 28.

Third-party platforms felt the ripple results too. Railway reported intermittent GitHub authentication failures and “Repo Not Discovered” errors from January 26-29 after hitting GitHub’s OAuth token fee limits.

GitHub says it is implementing stronger monitoring, improved take a look at environments, and tighter configuration safeguards following the January incidents. The corporate additionally flagged that February 9 incidents will seem in subsequent month’s report—early February already noticed AWS-related points impacting the platform.

For groups counting on GitHub Copilot as core infrastructure, the OpenAI dependency provides a variable exterior GitHub’s direct management. When your AI assistant’s upstream supplier stumbles, your builders wait.

Picture supply: Shutterstock