Skip to content

GitHub Copilot Statistics 2026: Users, Pricing, Productivity and Security

Matt Li By Matt Li Co-Founder and Director 10 min read
TL;DR: GitHub Copilot had 50 million users in July 2026, up from 15 million in April 2025, according to Microsoft. It had 4.7 million paid subscribers in January 2026. Since June 1, 2026, every plan runs on usage-based AI credits, with individual plans from $10 to $100 a month and business plans at $19 or $39 per user.

One in three pull requests on GitHub now involves an AI agent, Satya Nadella told investors on July 29, 2026. Agents are also why Copilot’s pricing changed: GitHub moved every plan to metered credits because agent work was driving up the cost of running Copilot.

Key takeaways
  1. 1Copilot now offers 30 model options from six providers, including Microsoft’s own MAI-Code.
  2. 2In GitHub’s own 2022 trial, Copilot users finished a task 55% faster. An independent 2025 trial found experienced developers 19% slower with AI tools.
  3. 329.5% of Python snippets Copilot wrote in public GitHub projects carried a known security weakness.
  4. 4Among developers who use AI agents, 67.9% have used Copilot, second only to ChatGPT.

How Many People Use GitHub Copilot?

GitHub Copilot had 50 million users in July 2026, Nadella said on Microsoft’s fourth-quarter earnings call. The count has more than tripled in 15 months: it stood at over 15 million in April 2025, 20 million that July and over 26 million in October.

Column chart of GitHub Copilot users as stated on Microsoft earnings calls: over 15 million in April 2025, 20 million in July 2025, over 26 million in October 2025 and 50 million in July 2026.

Microsoft counts free, student and paid accounts together, so the total measures reach rather than revenue. The paid figure is smaller and moves faster. Microsoft reported over 1.3 million paid Copilot subscribers in January 2024 and over 4.7 million in January 2026, up 75% in a year. Pro+ subscriptions for individual developers rose 77% in that one quarter.

~140,000
Organizations using Copilot, April 2026
60%+
Quarter-on-quarter Copilot revenue growth, April to June 2026
90%
Of the Fortune 100 using Copilot, July 2025
$2B
GitHub annual revenue run rate, July 2024
Source: Microsoft earnings calls, July 2024 to July 2026.

Enterprise subscribers nearly tripled in the year to April 2026, when Microsoft put the number of organizations at nearly 140,000. In January 2024 it had reported more than 50,000 organizations on Copilot Business alone.

Siemens moved to the full GitHub platform after rolling Copilot out to 30,000 of its developers. At AMD, tens of thousands of developers accept hundreds of thousands of lines of Copilot suggestions each month, Microsoft said in October 2025.

Copilot is also how many developers start on GitHub. In October 2025 Microsoft said 80% of new developers on GitHub use Copilot within their first week. Microsoft also said the launch of Copilot in VS Code drew more than a million sign-ups in its first week.

GitHub reported 225 million users in July 2026. In October 2025 it had described itself as home to over 180 million developers.

Revenue

Microsoft has not given Copilot revenue in dollars on its earnings calls. In July 2024 it said GitHub’s annual revenue run rate had reached $2 billion, and that Copilot drove over 40% of GitHub’s revenue growth that year.

Copilot was by then a larger business than all of GitHub was when Microsoft bought it in 2018. In the April to June 2026 quarter, which included the first month of usage-based billing, Copilot revenue grew more than 60% on the previous quarter.

How Much Does GitHub Copilot Cost in 2026?

Copilot costs $0 to $100 a month for individuals and $19 or $39 per user for companies. Since June 1, 2026, every plan includes a monthly allowance of GitHub AI Credits, and chat, agent mode, code review, the cloud agent and the CLI all draw them down. Code completions stay unlimited on paid plans.

PlanPrice a monthIncluded AI creditsNotes
Free$0Limited2,000 completions a month
Pro$10$15 ($10 base + $5 flex)Cloud agent, code review, Claude Code and Codex agents
Pro+$39$70 ($39 base + $31 flex)Premium models including Opus, audit logs
Max$100$200 ($100 base + $100 flex)Priority access to new models
Business$19 per user$19, pooled across the orgBudget controls, IP indemnity
Enterprise$39 per user$39, pooled across the orgTwice Business’s included usage
Source: GitHub Copilot plans page, September 11, 2026.

GitHub charges credits by tokens: input, output and cached tokens, at each model’s listed API rate. A long agent run on a frontier model therefore uses more of the month than a quick chat on a small one.

Grouped column chart of monthly GitHub AI Credits included per Copilot plan from June 1, 2026: Pro $10 base plus $5 flex, Business $19, Pro+ $39 base plus $31 flex, Enterprise $39, Max $100 base plus $100 flex.

Individual plans add a flex allotment on top of the base, which GitHub says may change as model prices change. Business and Enterprise get no flex, but their credits pool across the organization, so a light user’s unused allowance covers a heavy one.

Admins can cap spend at the enterprise, cost centre or user level. Existing business customers got higher promotional allowances of $30 and $70 for June to August 2026.

How the pricing changed

Oct 28, 2025
Agent HQ: GitHub opens Copilot to coding agents from OpenAI, Anthropic, Google, Cognition and xAI
Apr 27, 2026
Usage-based billing announced: AI credits to replace premium requests
May 12, 2026
Max plan: $100 a month, plus flex allotments on Pro and Pro+
Jun 1, 2026
AI credits go live on every plan

GitHub gave the reason in its April 27 announcement:

“Agentic usage is becoming the default, and it brings significantly higher compute and inference demands.”

GitHub, April 27, 2026

Microsoft’s earnings calls show the cost. On April 29 it said increased GitHub Copilot usage was pushing down Microsoft Cloud’s gross margin. On July 29 it said margins improved through the quarter after the June change to usage-based pricing.

Does GitHub Copilot Make Developers Faster?

In controlled tests, it depends on the task and the developer. GitHub’s own 2022 experiment gave 95 professional developers the same job, writing an HTTP server in JavaScript. The half with Copilot finished in 1 hour 11 minutes on average, 55% faster than the 2 hours 41 minutes it took the rest.

Proportional circles from GitHub's 2022 controlled experiment with 95 developers building an HTTP server in JavaScript: 71 minutes with Copilot against 161 minutes without, 55% faster.

More of the Copilot group also finished: 78%, against 70% without it. The task was a small, self-contained program.

Two later trials measured work on real codebases, and they point in different directions.

Accenture with GitHub, 2024
  • +8.69% pull requests per developer
  • +15% pull request merge rate
  • +84% successful builds
METR, 2025
  • 19% slower on 246 real issues
  • 16 experienced open-source developers
  • Tools: Cursor Pro with Claude 3.5 and 3.7 Sonnet

The Accenture study was a randomized trial of Copilot inside a large company. METR tested a different tool, Cursor, on large open-source projects the developers already knew well. Its most quoted finding is about perception: the developers expected AI to speed them up by 24%, and afterwards still believed it had sped them up by 20%.

How Developers Use GitHub Copilot

Once they have a licence, most developers start straight away. In the Accenture rollout, 81.4% installed the Copilot extension the day they got access, and 96% of those accepted a suggestion the same day. Two-thirds, 67%, used it at least five days a week.

  • About 30% of Copilot’s suggestions were accepted.
  • 88% of the characters Copilot generated stayed in the editor.
  • 90% of developers said they felt more fulfilled in their jobs, and 95% enjoyed coding more.
  • 43% found Copilot extremely easy to use, and 51% rated it extremely useful.

GitHub’s earlier survey of more than 2,000 developers pointed the same way: 73% said Copilot helped them stay in flow, and 87% said it saved mental effort on repetitive work.

Use has shifted from completions to agents. Developers merged more than 500 million pull requests on GitHub in the year to October 2025, and by July 2026 one in three pull requests involved an agent.

Microsoft said in April 2026 that usage of Copilot CLI, the terminal version, was nearly doubling month over month. By July it reported that millions of developers had used its own coding model, MAI-Code-1-Flash, inside Copilot, with 10% lower median token use.

The model list is now long. The plans page lists 30 model options from Anthropic, OpenAI, Google, xAI, Kimi and Microsoft, and each model draws credits at its own rate.

Where Copilot Ranks Among AI Coding Tools

Copilot is the most used coding-specific assistant among developers who work with AI agents. In the 2025 Stack Overflow Developer Survey, 67.9% of the 8,323 respondents who use or develop AI agents had used GitHub Copilot, behind only ChatGPT at 81.7%. Claude Code came fourth at 40.8%.

Bar chart from the 2025 Stack Overflow Developer Survey of AI assistants used by 8,323 respondents who use or develop AI agents: ChatGPT 81.7%, GitHub Copilot 67.9%, Google Gemini 47.4%, Claude Code 40.8%, Microsoft Copilot 31.3%, Perplexity 16.2%.

Trust lags use. In the same survey, more developers actively distrusted the accuracy of AI tools, 46%, than trusted it, 33%. Only 3% highly trusted the output. Use keeps rising regardless: 84% of all respondents used or planned to use AI tools, up from 76% a year earlier, and 51% of professional developers used them daily.

Copilot’s closest paid rival changed hands this summer. SpaceX completed its acquisition of Cursor on August 14, 2026, and our Cursor statistics cover that deal. Our Cursor vs GitHub Copilot comparison sets the two side by side, and Copilot’s Agent HQ now runs agents from Anthropic, OpenAI and others inside GitHub.

Security Risks in Copilot-Generated Code

About three in ten Python snippets that Copilot wrote in public projects contained a known weakness. Researchers led by Yujia Fu analyzed 733 Copilot-generated snippets found on GitHub, in a study accepted by ACM TOSEM. They found weaknesses in 29.5% of the Python snippets and 24.2% of the JavaScript ones, across 43 CWE categories.

Waffle chart showing that 29.5% of Python snippets generated by GitHub Copilot in public GitHub projects had at least one CWE security weakness, from a study of 733 snippets by Fu et al. in ACM TOSEM.

The same team found Copilot could help clean up after itself. Fed static-analysis warnings, Copilot Chat fixed up to 55.5% of the security issues.

Secrets are the other leak. GitGuardian’s State of Secrets Sprawl 2025 found 6.4% of public repositories using Copilot leaked at least one secret, a 40% higher exposure rate than the average public repository. It counted 23.8 million secrets leaked on public GitHub in 2024.

Our AI code review tools guide covers the scanners teams add on top, and our analysis of how much software AI writes tracks the share of code these tools now produce.

Hiring Developers Who Use Copilot Well

Second Talent tests every engineer on AI coding tools and matches you with pre-vetted developers from nine Asian markets within 24 hours. Tell us the role.

What’s your AI developer tool priority?
Select your situation below.
Pick an option above to get a tailored recommendation.
Build your AI development team
We test every engineer on how they use AI coding tools such as Copilot, Claude Code and Cursor, and we match you within 24 hours. Hire AI-ready developers →
Calculate your developer ROI
Copilot Business costs $19 per user a month. Salary is the bigger line. View Asia developer rates →
Expand your development capacity
Hire through our EOR service in Vietnam, the Philippines or Indonesia without an entity. Get EOR pricing →
Source niche technical talent
Need DevOps or cloud engineers? Tell us the role. Access specialist talent →

FAQs

Is GitHub Copilot free?

Yes, with limits. The Free plan includes 2,000 code completions a month, limited chat and agent use, and models such as Claude Haiku 4.5 and GPT-5 mini. Verified students get a separate Student plan.

What happens when I run out of AI credits?

On paid plans you can buy more at the published model rates, or an organization admin can cap spending so usage stops at the budget.

Hire developers who build with AI coding tools.

Pre-vetted senior engineers from Asia at 50 to 70% below US hiring costs, with first profiles in 24 hours.

Hire AI-native developers Apply as talent →
Matt Li

Written by

Matt Li is a tech-driven entrepreneur with deep expertise in global talent strategy, digital experience optimization, e-commerce, and Web3 innovation. He is the Co-Founder of Second Talent, a US-based company that connects businesses with top-tier tech professionals worldwide. Since launching the company in 2024, Matt has led its growth by leveraging technology to streamline remote hiring and scale distributed teams. With a background spanning product, operations, and innovation, Matt brings a cross-disciplinary perspective to the evolving digital economy. His work sits at the intersection of global talent, emerging technology, and scalable digital transformation.

More posts by Matt Li →

Loading available times…