According to a link post by developer Simon Willison, published 28 October 2026, GitHub released its annual Octoverse 2026 report at the GitHub Universe conference, where EVP Jay Parikh presented several headline numbers. Willison, who wrote that GitHub gave him a free ticket through its "GitHub Stars" program, reported the following figures as highlighted during the announcement. Merged pull requests tripled in less than a year to 220 million per month in 2026. Commits more than tripled to 7.38 billion per month, with most of that increase occurring in the second half of the year. The rate of new repositories doubled over the year to 30.3 million per month. Willison characterized the figures as a representation of the impact of agentic engineering on GitHub. The report also includes figures broken out by industry and a category covering repositories involving interoperability or reverse engineering. Per the methodology notes Willison cited, top countries by pushes to public repositories were adjusted by profile location using LLM classification to extract country codes, while public repositories were sorted into NAICS industry sectors using a fine-tuned multilingual embedding model drawing on name, description, topics and README. These figures come from a single secondary summary of GitHub's own report.
- Merged pull requests tripled to 220 million per month in 2026
- Commits more than tripled to 7.38 billion per month, mostly in H2
- New repository rate doubled to 30.3 million per month
- Figures attributed to GitHub EVP Jay Parikh at GitHub Universe, summarized by Simon Willison (28 Oct 2026)
What it means for you
GitHub's own data suggests a large jump in code activity this year, which it attributes to AI agents writing and merging code. For most organizations this is a weather report, not a to-do: it tells you the volume of AI-assisted code in the wild is climbing fast, but it doesn't change anything you run today. The practical takeaway is quieter — if your team ships software, more of the code flowing through your pipeline (yours and your vendors') is likely machine-generated, which raises the value of review and testing, not lowers it.
Try this
If your team writes or buys software, check one thing this week: does every change still pass through a human review and a test step before it reaches production? Rising automation only helps if that gate holds.
Who should care
Engineering leads, CTOs, and whoever owns code review and release process at organizations that build or heavily customize software.
Skip this if
You don't develop software in-house and don't manage a development team — this is industry trivia with no bearing on your operations.
Sources: Simon Willison — read the original