GitHub Outages: 52 Incidents in the First Nine Months of 2026, More Than Any Full Year in GitHub’s Own Reports

GitHub outages are getting more frequent by GitHub’s own count. Its monthly availability reports list 52 incidents from January to September 2026, up from 22 in the same months of 2025 and more than any full year since the reports began in mid-2020. Those incidents added up to 157 hours of degraded service, against 96 hours a year earlier.

Since July 2020 GitHub has published a GitHub Availability Report every month, describing each incident that resulted in degraded performance. We downloaded all 75 reports, extracted every itemized incident (184 after merging follow-ups that a later report repeats), and coded each for the services named and what set it off. As a second view, we pulled all 1,881 incidents from GitHub’s status page.

GitHub outages by month: the scorecard

Incidents itemized in GitHub’s monthly availability reports, by the month each began. Blank cells: before the first report (July 2020) or not yet reported (October 2026 onward).

The grid shows three phases. From mid-2020 to 2021, GitHub reported between zero and 3 incidents a month. A first wave came in spring 2023, with 14 incidents from March to May. Then 2026 broke the pattern: 10 incidents in Apr 2026 and at least four in every month from Feb to Aug. Our itemized count matches the total in GitHub’s own summary line in 63 of the 67 reports that state one; in the other 4, GitHub itemized one more incident than its summary counted (Codespaces incidents described separately in 2022, one added retroactively, and January 2025).

Is GitHub reliability getting worse?

On GitHub’s own numbers, yes, in frequency and in total time. January to September 2026 had 52 reported incidents against 22 in 2025 and 22 in 2024 over the same months. Total incident time rose from 96 to 157 hours. Incidents also ran longer: the median was 1 h 47 min in 2026 against 1 h 8 min in 2025.

The status page shows the same direction, with a caveat. It logged 243 incidents from January 1 to October 14, 2026, up 59% on the same period of 2025. In April 2026 GitHub committed to “statusing incidents both large and small” and added a new Degraded Performance level, which could inflate later counts. But the rise came before that change: the first quarter of 2026 had 94 incidents against 46 in the first quarter of 2025 (up 104%), and after April the monthly count fell from an average of 31 in January to March to 22 in May to September.

All incidents on githubstatus.com by the month they were opened (scheduled maintenance excluded). October 2026 runs to October 14.

*2026 runs to October 14. Impact labels are not comparable across 2026: the “critical” level first appears on 2026-04-22, after GitHub changed its severity scale.

What about the Azure migration?

GitHub has described the move in primary sources. In March 2026 its CTO wrote that 12.5% of GitHub traffic was served from the Azure Central US region; in August he wrote that “Azure serves roughly 58% of GitHub’s platform load and half of all Git operations, up from 12% of platform load in May.” Incidents rose over the same period, but the counts cannot separate the migration from the growth GitHub says it was migrating to absorb: in August it reported that monthly commits had grown from 1.4 billion to 2.9 billion since April. GitHub’s own explanation is load, not the move:

“Neither outage was caused by a code or configuration change. Both incidents were capacity failures at their core.”GitHub CTO Vlad Fedorov on the August 6 and August 17, 2026 outages, github.blog, August 20, 2026

Which GitHub services go down most?

GitHub Actions is named in more report incidents than any other service: 50 of 184. The big change in 2026 is Copilot, named in 25 of the 52 incidents this year against 16 in all of July 2020 to 2025. Copilot’s dependence on outside model providers shows up directly: GitHub added a separate “Copilot AI Model Providers” component to its status page in April 2026, and 17 status incidents have used it since.

Services named in each incident’s impact description (first two sentences, or the “Services impacted” line in newer reports). One incident can name several.

What causes GitHub outages?

We coded each incident by what GitHub says set it off. Across all 184, 92 (50%) followed a change GitHub made: a deployment, configuration change, database migration, credential rotation or maintenance step. Load and capacity problems account for 29, third parties and cloud providers for 25, infrastructure or hardware failures for 19, latent bugs and integer limits for 11, and expired certificates or credentials for 7.

The mix has moved only modestly. Change-triggered incidents were 44% of the total through 2023 and 50% in 2026. Third-party causes, including AI model providers behind Copilot and upstream cloud services, rose from 11% to 15%. Load and capacity incidents went from 14 through 2023 (19%) to 7 in the first nine months of 2026 (13%), so by trigger, load is not a growing share. That does not contradict GitHub’s account: several 2026 incidents coded as changes, such as the March 3 deploy that expired every user’s settings cache at once, did their damage through load, and the two capacity failures GitHub highlighted in August were among the longest of the year.

Every incident in GitHub’s availability reports

Newest first. Start time and duration as GitHub’s report gives them; July 2026 entries have no start time in the report.

What this means for MSPs

Budget for more GitHub downtime, not less. 157 hours of reported incidents in nine months is already more than any earlier full year except 2024 (160 hours, of which 76 came from one email delay). For clients whose deployments, CI or code review run through GitHub, write down what happens when Actions or Git operations stop for a working afternoon, and keep a way to ship an urgent fix without GitHub-hosted runners. If a client contract promises an uptime figure that depends on GitHub, check it against our 99.9% uptime calculator.

Treat Actions and Copilot as separate risks. Actions is the most often named service. Copilot incidents now often start at an outside model provider, and GitHub’s own advice in its April status update is that many Copilot features let users pick another model when one is down. Show client developers how to switch models or fall back to working without the assistant.

Subscribe to the feed, and do not rely on Downdetector alone. githubstatus.com offers email subscriptions and a JSON feed. Pipe the components your clients use into your PSA. Cloud dependencies fail in clusters; our record of AWS outage history and our anthology of Cloudflare outage post-mortems show the same habit of one shared component taking several services down.

Frequently asked questions

Is GitHub down right now?

This page is a history, not a live monitor. Check githubstatus.com, which lists open incidents for each service, or its JSON feed at /api/v2/summary.json. GitHub has said it now posts incidents both large and small there, so an issue that is not listed is more likely on your side.

Why is GitHub having so many outages?

GitHub’s CTO said in March 2026 that the instability was driven mainly by rapid load growth, coupling between services and an inability to shed load from misbehaving clients, and in August that the two worst outages of that month were capacity failures. In GitHub’s own reports, 7 of 52 incidents in 2026 were load or capacity problems, but changes such as deployments and configuration updates still set off 26.

How many outages has GitHub had in 2026?

GitHub’s monthly availability reports list 52 incidents from January to September 2026, against 22 in the same months of 2025 and 34 in all of 2025. Its status page, which also lists smaller issues, logged 243 incidents from January 1 to October 14, 2026.

What caused the GitHub outage on August 17, 2026?

GitHub said traffic reached a new peak and a critical infrastructure component in its Central US data center failed to scale with it; the pressure spread and caused authentication failures across services. Its availability report gives the incident as 7 h 35 min from 13:40 UTC, and its CTO’s post put the outage at 7 hours and 47 minutes.

Did moving GitHub to Azure cause the outages?

The counts cannot show that. GitHub said 12.5% of its traffic ran from Azure in March 2026 and about 58% of platform load by August, while incidents were already rising in 2025 and kept rising as the move went on. GitHub’s own explanations point to load growth and capacity, and it describes Azure as the way it is adding capacity.

How long do GitHub outages last?

The median incident in GitHub’s availability reports lasted 1 h 32 min across all 184 incidents since July 2020, and 1 h 47 min in 2026. The longest was the April 11 to 14, 2024 email delay, at 3 d 4 h 23 min.

Methodology

References

1. GitHub Blog, GitHub Availability Report series (75 monthly reports, 2020-08 to 2026-10). Accessed October 2026.
2. GitHub Status, incident history and per-incident JSON. Accessed October 2026.
3. GitHub Blog, “Addressing GitHub’s recent availability issues,” March 11, 2026. Accessed October 2026.
4. GitHub Blog, “Bringing more transparency to GitHub’s status page,” April 17, 2026. Accessed October 2026.
5. GitHub Blog, “An update on GitHub availability,” April 28, 2026. Accessed October 2026.
6. GitHub Blog, “The August 17 outage, and the work ahead,” August 20, 2026. Accessed October 2026.
7. GitHub Blog, “GitHub availability report: September 2026” (latest report used). Accessed October 2026.