San Francisco, 28 October. Developers start more than 40% of pull requests on GitHub with artificial intelligence agents, up from 17% in March 2026, according to the platform’s annual Octoverse report; over 12 months code was pushed to GitHub 17.9 billion times — more than in 2024 and 2025 combined.
Commits — recorded code changes — across public and private repositories reached 34.3 billion in the year, more than three times the level of two years ago. Merged pull requests rose to 1.3 billion, an average of 3.5 million a day. Among repositories created this year, a project with an agent configuration file logged on average eight times as many commits as a project without one. Since June about 55% of all commits came from brand-new repositories. GitHub Actions automation ran a job 1,440 times a second, three times as often as a year earlier.
GitHub describes the change this way: for a growing share of work the first move is now directing an agent, and review is how the developer shapes what comes back. In September 61% of GitHub Copilot users worked with more than one AI model. To cope with the volume, the company, according to Jay Parikh, head of CoreAI at Microsoft, is not only adding capacity but also changing its architecture, moving the service to a multi-region deployment in the Azure cloud; at the GitHub Universe conference it showed Copilot Fusion, which picks a model for each task, and the deeper Copilot code review Max.
The platform now serves more than 225 million developers. The largest communities are in the United States (38.3 million) and India (32 million), each of which added more than 10 million in a year; Brazil was third by growth (3.28 million). Public contributions to open source reached 2.5 billion in the year, more than double 1.12 billion in 2025. The publication contains no data on Kazakhstan or Central Asia.
What it means: On the largest code platform, programming has shifted within six months from writing lines to setting tasks for agents and reviewing their work; the volume of code is growing faster than the number of people who write it.
Who benefits: GitHub and its owner Microsoft gain higher usage and demand for paid Copilot features. Companies gain faster development. GitHub itself acknowledges the flip side: more code means more requests for review, and speed only helps if the quality of what ships is maintained.
Background: GitHub is the world’s largest service for hosting code and collaborative development and has belonged to Microsoft since 2018. An AI agent, unlike an autocomplete assistant, carries out a multi-step task itself: it reads code, makes changes, runs checks and opens a pull request. Retellings of the report on Telegram cited “7.4 billion commits in September” and “4 million processor cores”; GitHub’s texts contain neither: 7.4 million is the forecast number of monthly contributors to JavaScript projects by September 2027.
