ItInsightsWorld General: Inside Octoverse 2026 as Agents, Models, and Builders Redefine GitHub Work
This ItInsightsWorld General reading of GitHub Octoverse 2026 treats the report as a three-layer story: agents are beginning more work, model choice is spreading, and builders are entering from more industries. The headline volume is huge: across GitHub, developers pushed code 17.9 billion times in the past 12 months, a level the report says is above the combined 2024 and 2025 totals. The report also carries smaller signals, such as a Nepal flood bulletin on GitHub Pages that grew into Gati, spacecraft software work at NASA and the Jet Propulsion Laboratory, and open-source firmware that let people customize e-readers.
Headline signals in one view
- Agent-aware repositories moved differently. New repositories that carried an agent configuration file averaged eight times as many commits as new repositories that did not.
- Agent initiation rose during the year. In September, agent-started pull requests passed more than 40%, after March recorded 17%.
- Model selection became plural. In September, 61% of GitHub Copilot users who selected models themselves worked with more than one model.
- Industry coverage widened. For the first time, public repositories are viewed by sector, and repositories serving mining, construction, health care, and agriculture each grew several times over since 2022.
- Growth outran a five-year projection. One year into earlier projections, every country had already closed at least a fifth of its distance to 2030.
Activity, automation, and staying power
The year's pace shows up in several layers. Public and private commits reached 34.3 billion, more than three times the level from two years ago. Merged pull requests climbed to 1.3 billion, averaging 3.5 million a day.
Automation also expanded. GitHub Actions ran a job 1,440 times every second, which equals 124 million times a day and is more than three times the prior year total. Dependabot pull requests merged more than 30 million times, about double 2025.
The report's telemetry shows a stepwise pattern: activity climbs from January, strengthens around June, and remains elevated. It places that shift near improvements in models, the surrounding harnesses, and agents that could persist across days, coordinate other agents, and run work remotely as well as locally.
Participation keeps widening
New repositories account for much of the recent output: from June onward, about 55% of commits came from newly created repositories. At the same time, long-standing developers stayed active. Those who pushed code in the past 12 months had been on GitHub for an average of 36 months, and more than 75% of developers who opened a pull request in August pushed code again in September. The average developer also spread across 10% more repositories than a year ago.
Open-source contribution grew sharply. Public contributions reached 2.5 billion in 2026, more than double the 1.12 billion total from 2025. The report names @kinlane, with more than 1 million contributions and agentic workflows used to keep a machine-readable public API catalog current; @kim-em, with more than 178,000 contributions while maintaining mathlib; and @andig, an evcc maintainer with over 21,000 contributions.
Community growth across countries
GitHub now serves more than 225 million developers. The two largest national communities are the United States, at 38.3 million, and India, at 32 million; each added more than 10 million developers over the year. Brazil added 3.28 million developers, ranking third among national additions.
| Country | Total developers |
|---|---|
| United States | 38.3 M |
| India | 32.0 M |
| China | 12.4 M |
| Brazil | 10.2 M |
| Japan | 7.15 M |
| United Kingdom | 6.47 M |
| Indonesia | 6.34 M |
| Germany | 6.30 M |
| Russia | 5.07 M |
| France | 4.75 M |
Projection pressure and regional drivers
Earlier projections once suggested India would reach 57.5 million developers by 2030 and the United States 54.7 million. One year into that five-year window, every projected country has closed at least a fifth of the distance to its own 2030 number, with the United States at 38.6% and India at 28.3%.
| Region | Reported driver | Examples |
|---|---|---|
| Americas | Fintech and digital payments |
|
| EMEA | AI adoption in manufacturing and startups, plus fast-growing communities in Africa |
|
| APAC | Government investment in AI infrastructure and talent |
|
Forecasting becomes less certain
The report runs more than 40 forecasting models forward to 2031, using historical sign-up and usage data. About half call for extreme growth, with developer counts on GitHub more than tripling. Just under half call for substantial growth, described as doubling to tripling. Only three models point to a path of under 2x growth.
The spread matters because the five years beyond the report are harder to pin down. What remains stable is not a single number, but a community that is growing everywhere.
Building is happening everywhere
Earlier analyses used the phrase everywhere to describe more developers in more countries. The 2026 view adds more industries, and more fields are visible in public repository activity.
| Rank | Country |
|---|---|
| 1 | United States |
| 2 | India |
| 3 | China |
| 4 | Brazil |
| 5 | Germany |
| 6 | United Kingdom |
| 7 | Korea |
| 8 | France |
| 9 | Japan |
| 10 | Canada |
Industry classification shows new builders
The Octoverse team uses the North American Industry Classification System, or NAICS, to sort public repositories by the sector they serve. A language model classified samples created from 2022 to 2026 based on name, description, topics, and README. The analysis tracks two signals: whether a repository mentions AI, and whether an agent opened a pull request against it.
The signals do not always align. In administrative and support services, mining, health care, and agriculture, more than four in 10 repositories mention AI, while agent-opened pull requests are rare at less than 3 percent. Construction differs: about a quarter mention AI, and about 9 percent have agent-opened pull requests, the highest sector share, with manufacturing next.
Since 2022, sectors outside software have climbed. Professional, scientific, and technical services moved from fourth to second among industries, behind information. Agriculture, forestry, fishing and hunting rose two spots, while finance and insurance and health care and social assistance each moved up one. Compared with their own 2022 baselines, mining, quarrying, and oil and gas extraction grew roughly 65 times; administrative and support services and construction grew 17 to 18 times; and even manufacturing, the slowest group mentioned, almost tripled.
Professional, scientific, and technical services, which grew about 5 times, is the work of researching, engineering, and advising. In biochemistry, the work behind AlphaFold helped give Demis Hassabis and John Jumper a share of the 2024 Nobel Prize in Chemistry. In computer science, Tensor2Tensor accompanied the paper that introduced the Transformer, the architecture behind today's large language models, Attention Is All You Need. In medicine, sars-cov-2-stability held code and data behind an early paper on virus survival in air and on surfaces. In physics, eht-imaging was one of two open-source libraries used to reconstruct the first image of a black hole.
Field-level builders use agents
These sector shifts come from people applying agents to familiar problems. A manufacturing team turned a photo of tools into a custom-fit, 3D-printable drawer organizer. A health-care research lab helped scientists analyze huge digital tissue-sample scans, the kind pathologists study to understand disease. In Copenhagen, a university lab classified video of chickens playing, from frolicking to worm chasing. The University of Waikato energy center built a toolkit for industrial plants to find reusable heat, and Bulgaria's digital ministry open-sourced a portal that lets anyone trace who won a public contract and for how much. A novelist also built a free app that makes a manuscript look like a book while writing.
Agents reshape pull requests and review
Developers increasingly open pull requests by specifying work and letting agents draft it. In September 2026, more than 40% of pull requests started with an agent's first draft, up from 17% in March. That volume reached 110 million in September alone, about 6x March's total.
Agent-started pull requests also tend to touch multiple files: about 87%, compared with 79% for pull requests without detected agent involvement. Merge speed changed too. In February, 83% of agent-involved pull requests merged within two weeks, while the non-agent comparison group sat at roughly 86%. By September, agent pull requests merged within two weeks at 90%, while the comparison group was about 86%.
Agentic usage within GitHub Copilot is broadening as well: 55% of monthly active users now use agentic features, up from 35% a year ago. The most independent mode, Copilot cloud agent, grew 230% year over year, compared with 85% for agent chat. At the same time, public repositories mentioning interoperability or reverse engineering grew more than 300% year over year, from about 30,000 created in 2025 to more than 130,000 in 2026.
Review tools are catching up. Stacked pull requests launched as a preview on July 30, and more than 264,000 developers had used them by September. AI-assisted code review and GitHub Advanced Security grew more than 90% year over year. The platform also screens exposed secrets through push protection, which ran 5.8 billion scans this year, nearly four times 2025.
Model portfolios move quickly
GitHub Copilot users who pick models themselves increasingly switch among options. In September 2026, 61% used more than one model across sessions, up from 50% four months earlier.
The available set expanded: paid GitHub Copilot users used 29 distinct models in May and 47 in September, and no single model carried more than 25% of requests that month. Some decisions reflect cost and provider familiarity.
When explicitly selected requests are counted, excluding auto model selection, GitHub and Microsoft staff usage, background utility models, duplicate monitoring and testing requests, and requests made with a developer's own API key, a 100-request split falls at 38 general purpose models, 37 deep reasoning and planning models, and 25 fast and lightweight models. That leaves about 25% of selected traffic moving through a fast model.
Viewed over the past three months, three patterns appear. New models can claim a larger request share than the share of developers who have tried them, which suggests early adopters use them heavily. When request share declines, the share of developers still using a model declines more slowly. Requests also spread across all three model types, so many developers blend capabilities rather than staying with one.
Some developers also let the platform decide: 63% of paying GitHub Copilot developers used auto model selection at least once in September 2026. Project HydraFusion, a research preview of model orchestration, routed almost 60,000 sessions in the CLI and the GitHub Copilot app. It is now generally available as GitHub Copilot Fusion, which combines model orchestration with auto model selection and chooses a model behind the scenes.
Language choices gain a new influence
Language choice has usually followed what teams want to build, existing systems, and known skills. Octoverse 2026 asks whether a fourth influence is emerging: the language an agent picks when a developer does not specify one.
The 2026 order still starts with TypeScript, then Python, then JavaScript. From September 2025 to September 2026, TypeScript added 3.62 million monthly contributors, Python added 3.23 million, and JavaScript added 1.82 million. Those were the largest gains among the top 10 languages. Rust grew fastest by percentage among the top languages, doubling in the past year.
Extending recent trends, TypeScript and Python reach near 13 million monthly contributors each by September 2027, JavaScript reaches 7.4 million, and every other language in the top 10 is projected at about 2.2 million.
In the report's language chart, the detected-AI bands include about 40% to 44%, about 21% and 34%, more than one in five, and 24%.
What the 2026 pattern suggests
For ItInsightsWorld General readers, the Octoverse 2026 lesson is less about a single number and more about the interaction of agents, models, and industry participation. The workflow layer is changing fastest: agent-started pull requests are common, review tooling is expanding, and model choice is becoming a regular engineering decision rather than a default setting.