GitHub Statistics 2026

Picture the budget memo due Friday. It needs one line on how many developers use GitHub and another on how much code moves through it. The first number you find comes from a 2025 blog post, and the next from an aggregator that names no source. Neither says which year it describes.

The short answer: GitHub says it now serves more than 225 million developers, a figure Microsoft repeated on its FY26 Q4 earnings call in July 2026. For the year to 30 September 2026, GitHub reports 34.3 billion commits and 17.9 billion pushes in its Octoverse 2026 report. Each figure below carries its period, population and evidence type so you can cite it cleanly.

Key GitHub statistics for 2026

• 225 million+ developers: GitHub "serves" this many accounts worldwide, per Octoverse 2026, published 28 October 2026. Microsoft's CEO gave the same figure on the FY26 Q4 call, 29 July 2026 (Microsoft).

• 36 million+ new accounts: created from 1 September 2024 to 31 August 2025, up 23% year on year, per Octoverse 2025.

• 34.3 billion commits: all repository types, public and private, 1 October 2025 to 30 September 2026. GitHub says this is "more than three times where they stood two years ago" on its recalculated basis (Octoverse 2026).

• 17.9 billion pushes: same window and source. A push can carry several commits, so pushes and commits are different units.

• 1.3 billion merged pull requests: same window, averaging 3.5 million a day (Octoverse 2026).

• 124 million GitHub Actions jobs a day: GitHub's equivalent of 1,440 job starts every second over the same window, more than three times the prior year (Octoverse 2026).

• More than 40% of pull requests started with an agent's first draft by September 2026, up from 17% in March 2026. This counts only agent activity GitHub can detect (Octoverse 2026).

• 630 million repositories: public and private, cumulative as of GitHub's 2025 report, which ran to 31 August 2025. No first-party 2026 total has been published (Octoverse 2025).

• 50 million GitHub Copilot users: disclosed by Microsoft on 29 July 2026. Microsoft does not define the figure as monthly active users, paid seats or unique people (Microsoft FY26 Q4 call).

• 60 million Copilot code reviews: passed as of GitHub's 5 March 2026 post, accounting for more than one in five code reviews on GitHub (GitHub Blog).

• 28.65 million new hardcoded secrets added to public GitHub commits in 2025, a 34% rise, per GitGuardian's own scan in The State of Secrets Sprawl 2026, published 17 March 2026.

• 69.5% of respondents used GitHub public projects considerably or consistently in the past year, in the 2026 Stack Overflow Developer Survey (7,460 of 10,741 respondents). See Stack Overflow's community data for the question wording.

Get the Mobile Testing Playbook Used by 800+ QA Teams

Discover 50+ battle-tested strategies to catch critical bugs before production and ship 5-star apps faster.

Email addressGet Instant Access

100% Free. No spam. Unsubscribe anytime.

How many developers use GitHub?

GitHub's headline verb is "serves", and it is the right one. The 225 million figure counts accounts, not people who did anything in a given month. Octoverse 2026 is the first-party source, and Microsoft's FY26 Q4 call, three months earlier, said "GitHub now has 225 million users". Two first-party statements that agree make the figure safe to publish.

The prior report gave a lower base: 180 million-plus on the same vendor basis, for the year to August 2025. The windows and definitions differ, so do not subtract one from the other to get a growth rate.

GitHub also publishes a country breakdown labelled September 2026. The United States has 38.3 million developer accounts, India 32.0 million and Brazil 10.2 million. Brazil added 3.28 million in the year, and GitHub says the US and India each added more than 10 million. Geography follows self-reported location, and GitHub lists only countries with at least 100 developers.

Do not turn these numbers into "active developers", because GitHub's own definitions show why. Octoverse counts the distinct accounts that took part each month through pushes, pull requests, issues, reviews or comments, attributed to a repository's primary language, and warns that an account "can count in several languages". GitHub Innovation Graph defines developers differently: accounts located in an economy based on mode daily location.

Neither is a global count of active developers, and the per-language figures cannot be added together. None of the primary sources cited here publishes a monthly-active-developer total for GitHub.

How much activity does GitHub handle?

Octoverse 2026 measured 1 October 2025 to 30 September 2026, inclusive, counting public and private repositories unless it says otherwise. It reports 17.9 billion pushes, "more than 2024 and 2025 combined", and 34.3 billion commits. It reports 1.3 billion merged pull requests, which "more than tripled" over the same two-year comparison, so read that as two years, not one.

Dependabot merged more than 30 million pull requests, about double the 2025 pace. GitHub logged 5.8 billion push-protection scans, nearly four times the 2025 pace. Public contributions reached 2.5 billion, against 1.12 billion in 2025.

GitHub defines the units itself: "A push sends one or more commits to GitHub. Contributions count each push once, so commit totals can exceed contribution totals. Actions are counted in jobs, not minutes." If your memo sets this year's commit total beside last year's "commits pushed" figure, it compares two different things. Octoverse 2025's "nearly 1 billion commits pushed" covered September 2024 to August 2025 on the older basis.

How many repositories are on GitHub?

Three repository numbers circulate, and each answers a different question.

Octoverse 2025 reported 121 million repositories created in 2025, a record, including 58 million new private repositories (up 33%) and more than 230 new repositories every minute, an average for the year rather than a live rate. It put 63% of all repositories as open source or public, and 81.5% of 2025 contributions as happening in private repositories.

Do not present 630 million as a current total; it is GitHub's 2025 figure. The GitCharts number is a beta count from a dashboard, tracking since it began, so it is not comparable with GitHub's cumulative count. The StackScan figure was still incomplete when collected.

If you follow a single project rather than the platform, the Appium statistics roundup shows how GitHub activity looks for one open-source testing framework.

What the language data shows

GitHub and independent crawlers rank languages on different bases, and each is right on its own terms.

On GitHub's measure, TypeScript stayed first by monthly contributors in 2026, with Python second and JavaScript third. TypeScript first overtook both in August 2025. Between September 2025 and September 2026, TypeScript added 3.62 million monthly contributors, Python 3.23 million and JavaScript 1.82 million. Rust roughly doubled, the fastest percentage growth among the top languages (Octoverse 2026).

A gain is a net change in a per-language count of accounts, not 3.62 million new people. Because the counts overlap, they cannot be summed into a platform total. GitHub also forecasts that TypeScript and Python will each reach about 13 million monthly contributors by September 2027. That is a projection, not a 2026 count.

The popular-repository stock tells a different story. StackScan's crawl of 480,428 public non-fork repositories with 100 or more stars, collected 24 September 2026, found Python as the primary language of 19.8%. JavaScript had 11.8%, TypeScript 7.6%, Java 5.5% and C++ 5.3%. TypeScript's share of newly created popular repositories rose from 1.2% in the 2012 cohort to 18.6% in 2026. A tracked set on GitCharts puts Python at 19.17%, excluding "Other" and repositories with no declared language; it does not state its coverage, so treat it as one dashboard's view.

So TypeScript leads on active contributor headcount, and Python leads on the stock of popular and tracked repositories. Both hold at once.

AI involvement also varies by language. Among repositories created in 2026 with an observed agent signal, TypeScript was the primary language in 35%, against 24% without one. For older repositories the gap was 27% against 17%. Detected AI involvement appeared in about 40–44% of merged pull requests in TypeScript, Python and Rust, and 21–34% in Java, C# and C++. These are detected signals, not a census of AI-written code.

Agents now start a large share of pull requests

By September 2026, more than 40% of pull requests on GitHub had started with an agent's first draft, up from 17% in March. That is about 110 million agent-started pull requests in September alone, roughly six times March's volume (Octoverse 2026).

Agent-started pull requests were about 87% multi-file, against 79% without detected agent involvement. In February, 83% of agent-involved pull requests merged within two weeks, against about 86% without; that share rose to 90% by September.

New repositories created in 2026 with an agent configuration file logged eight times as many commits as the average new repository without one. That compares two averages. It does not show that an agent caused the extra work, and GitHub's glossary notes that repositories without a configuration file may still use AI.

GitHub counts an agent-started pull request only when "a coding agent that we can detect" produced the first draft, and calls the agent signal "a floor, since not all agent use may be detected". Treat every agent figure here as detected and correlational, never as a productivity measure.

GitHub Copilot statistics

Microsoft discloses GitHub Copilot figures on its earnings calls. Three matter most, and each is dated and defined differently.

These are historical disclosures, not October 2026 counts. On the Q4 call Microsoft said Copilot revenue accelerated more than 60% quarter on quarter. That is a growth rate; no absolute GitHub Copilot revenue appears in these calls, so do not convert it into a revenue figure. The 30 million Microsoft 365 Copilot seats sometimes quoted alongside these belong to a different product.

Octoverse 2026 gives the closest GitHub comes to usage data. Among Copilot's monthly active users, 55% used agentic features in the year, up from 35% a year earlier. Copilot cloud agent usage grew 230% year on year, against 85% for agent chat. These are shares, not counts.

On model choice, 61% of Copilot users who selected a model themselves used more than one in September 2026, up from 50% four months earlier. Paid-plan developers used 29 distinct models in May and 47 in September, and no single model carried more than 25% of requests that month. Across developer-selected requests in the seven days ending 13 October 2026, 38 in every 100 went to general-purpose models, 37 to deep-reasoning and planning models and 25 to fast, lightweight models. That window excludes automatic model selection, GitHub and Microsoft staff usage and bring-your-own-key requests.

For code review, GitHub reported in March 2026 that Copilot had passed 60 million reviews, with usage up tenfold since its April 2025 launch. It accounted for more than one in five code reviews on GitHub, and more than 12,000 organizations ran it automatically on every pull request. Those figures are as of that post.

Octoverse 2025 added one adoption measure: nearly 80% of new GitHub developers used Copilot within their first week. That is first use, not retained or monthly use.

How healthy are GitHub's repositories?

StackScan's crawl describes a narrower and more useful slice for judging quality: 480,428 public non-fork repositories with at least 100 stars, collected 24 September 2026. Every figure below measures that sample, not GitHub as a whole.

Of these popular repositories, 38.5% (185,154) had no push in three years, and only 19.1% had been pushed to in the previous 30 days. Archived repositories made up 8.4%. For every archived repository, 3.9 others had gone three years without a push and were never archived, so quiet decay is far more common than declared abandonment.

On licences, 26.8% (128,584) had no licence GitHub could detect, which is not the same as being unlicensed in law. Among detected licences, MIT was most common at 35.8%, Apache-2.0 at 12.4% and the copyleft family at 10.2%. GitCharts reports a 39.60% MIT share, but for tracked repositories with an unstated denominator, so the two figures should never be averaged.

The 100-star floor excludes most repositories on the platform, including nearly all personal and early-stage projects. Read these figures as a picture of the popular tail, not the long tail.

How many secrets leak into public GitHub?

GitHub does not publish figures on leaked credentials across its public code. The most detailed measurement comes from GitGuardian, a secrets-detection vendor with a commercial interest in the findings, in its fifth annual State of Secrets Sprawl report, published 17 March 2026. Cite these as GitGuardian's own scan results.

• 28.65 million new hardcoded secrets were added to public GitHub commits in 2025, a 34% rise that GitGuardian calls the largest single-year jump it has recorded.

• 1,275,105 AI-service secrets were detected, up 81%. Eight of the ten fastest-growing detector types were tied to AI services.

• Commits assisted by Claude Code showed a 3.2% secret-leak rate, against a 1.5% baseline across public GitHub commits. GitGuardian cautions this "should not be read as a simple tool failure", because developers control what they push.

• Of valid secrets first found in 2022, 64% were still active and exploitable when retested in January 2026. This is a single-cohort figure.

• Validation-only prioritisation missed 46% of critical secrets, according to GitGuardian.

• About 28% of incidents originated entirely outside repositories, and leaks outside code were 13 percentage points more likely to be classified critical.

How developers say they use GitHub

Stack Overflow's Developer Survey is a self-reported, opt-in sample, so quote its percentages with their denominators and never call it a census.

In the 2025 survey, asked which collaborative work management and code documentation tools they had used regularly over the past year, 81.1% of all respondents (n = 30,065, multiple answers allowed) named GitHub. Jira had 46.4%, GitLab 35.6% and Confluence 32.8%. Among professional developers, GitHub reached 80.5% (Stack Overflow 2025). For the same tool category, GitHub scored 59.3% desired and 70.1% admired, where admired means people who used it in the past year and want to keep using it.

The 2026 survey asked a different question: 69.5% of respondents (7,460 of 10,741) used GitHub public projects considerably or consistently in the past year, with Stack Overflow itself at 68.6%. The wording is different and the respondent pool smaller than in 2025, so this is not a continuation of the 2025 community figure and cannot be compared with the 81.1% collaboration-tool result. Report 69.5% as its own result.

What these statistics don't measure

GitHub reports 34.3 billion commits, 1.3 billion merged pull requests and 124 million CI jobs a day. None of the sources cited here reports which of those changes broke something, how long a break took to surface, or whether the same failure would appear on an emulator as on a handset. GitHub's reports measure volume, adoption and merge speed rather than defects introduced. GitGuardian counts leaked secrets, not broken builds, and the Actions figure counts jobs, not outcomes.

Settling it would take a defined sample of repositories, a fixed test suite, real devices and emulators run side by side, and a reported time to detection. That is a study, not a statistic, which is why the public record does not carry one. Quash has specified a study along these lines but has not run it, so this article cites no Quash result, and no first-party Quash data covers GitHub activity.

Two more gaps matter for planning. No platform-wide monthly-active-developer count is published in the primary sources cited here.

The other gap is that GitHub and GitGuardian publish activity figures for overlapping periods that do not reconcile. GitHub reports nearly 1 billion commits pushed from September 2024 to August 2025, while its companion post titles the same 986 million figure "code pushes". GitGuardian reports about 1.94 billion public GitHub commits in calendar 2025, up 43%. The windows overlap only partly and neither publisher has reconciled the two, so this article calculates no ratio between them.

Methodology and sources

GitHub's 2026 figures cover 1 October 2025 to 30 September 2026, inclusive. GitHub recalculated its 2024 and 2025 figures on the same definitions and window, so compare recalculated numbers only with recalculated numbers. The 2025 report used 1 September 2024 to 31 August 2025.

Unit definitions follow GitHub's usage: pushes, commits and contributions are different counts, and Actions are counted in jobs. Per-language contributor counts overlap and are not additive.

Each publisher keeps its own definitions. Microsoft's Copilot headcounts are not defined as monthly active users or paid seats. StackScan's figures cover public non-fork repositories with at least 100 stars and are third-party measurements. GitCharts' repository total is a BETA count of tracked repositories. GitGuardian's findings come from its own scanning, and Stack Overflow's are self-reported survey results.

Conclusion

If you need one figure for a deck, use GitHub's 225 million, with its date and the word "serves". For activity, use 34.3 billion commits or 17.9 billion pushes and name the window. Avoid any single "GitHub statistics" total that mixes accounts, contributors, repositories and pushes.

Decide which question your piece answers before you pick a number. Developer reach, code volume, repository health, security exposure and Copilot adoption each have their own source and limits. Use the source that matches your question, and state its population and date in the sentence that uses it.