OpenAI’s Codex crossed 7 million weekly users on July 13, 2026, then climbed to 8 million within days after GPT-5.6 launched and Codex merged into the ChatGPT desktop app.
That’s up from fewer than 1 million in February 2026, one of the fastest scaling curves in software history.
Codex is OpenAI’s cloud-based software engineering agent, built to handle feature writing, bug fixing, refactors, and pull requests across sandboxes and developer workflows.
Codex is part of a broader wave of AI developer tools reshaping how software is built. If you’re exploring this space, see our breakdown of AI coding statistics and adoption trends.
This article covers Codex users, adoption, pricing, enterprise traction, benchmark performance, competitive context, and the broader AI coding tools market.
Codex AI Statistics 2026: Key Insights & Takeaways
- Codex crossed 7 million weekly users on July 13, 2026, then 8 million within days as GPT-5.6 launched and Codex merged into the ChatGPT desktop app.
- Combined weekly actives across Codex and the newly launched ChatGPT Work agent reached ~10 million by July 21, 2026, per OpenAI.
- Growth has been roughly 7-8x in five months: under 1 million in February to 7-8 million by mid-July 2026.
- Knowledge workers made up ~20% of Codex users as of June 2026 and were adopting it 3x faster than developers.
- Codex CLI, the open-source terminal agent, passed 99,624 GitHub stars and grew npm downloads from 82,000 (April 2025) to 41.8 million (May 2026).
- OpenAI’s broader developer platform has 4 million developers and processed 6 billion tokens per minute as of DevDay 2025.
- OpenAI had 1 million+ business customers and 7 million+ ChatGPT for Work seats as of late 2025.
Codex AI Statistics 2026 At A Glance
| Metric | Figure |
|---|---|
| Codex weekly users (July 13, 2026) | 7M+ |
| Codex weekly users (~July 20, 2026) | 8M+ |
| Combined Codex + ChatGPT Work actives (July 21, 2026) | ~10M |
| Codex weekly users (June 2, 2026) | 5M+ |
| Codex weekly developers (April 21, 2026) | 4M+ |
| Codex weekly users (February 2026) | <1M |
| Knowledge workers, share of users (June 2026) | ~20% |
| Codex CLI GitHub stars (July 19, 2026) | 99,624 |
| Codex CLI npm downloads (May 2026, from Apr 2025) | 41.8M (from 82K) |
| GPT-5.3-Codex SWE-Bench Pro (last public benchmark) | 56.8% |
| GPT-5.3-Codex Terminal-Bench 2.0 | 77.3% |
| GPT-5.3-Codex OSWorld-Verified | 64.7% |
| Codex Free plan | $0/month |
| Codex Go plan | $8/month |
| Codex Plus plan | $20/month |
| Codex Pro plan | From $100/month |
| OpenAI developers on platform | 4M |
| OpenAI API throughput (DevDay 2025) | 6B tokens/minute |
| OpenAI business customers | 1M+ |
| ChatGPT for Work seats | 7M+ |
| OpenAI ARR (2025) | $20B+ |
| OpenAI monthly revenue (March 2026) | $2B |
| AI code assistants market size (2025) | $8.51B |
| AI code assistants market forecast (2033) | $42.90B |
Codex AI Statistics: Deep Dive
1. User Statistics
Codex’s public user story is unusually transparent for a coding product, though it’s measured in platform-level milestones rather than a traditional consumer MAU figure.
| Date | Codex Weekly Active Users |
|---|---|
| February 2026 | <1M |
| March 5, 2026 | ~2M |
| April 8, 2026 | ~3M |
| April 21, 2026 | 4M+ |
| May 31 / June 2, 2026 | 5M+ |
| July 12, 2026 | 6M |
| July 13, 2026 | 7M |
| ~July 20, 2026 | 8M |
On July 9, 2026, OpenAI shipped GPT-5.6, merged Codex into the redesigned ChatGPT desktop app, and launched ChatGPT Work, a new agentic mode for knowledge workers, while beginning to retire its standalone Atlas browser.
Sam Altman said combined usage of Codex and ChatGPT Work rose 2.5x in the week following the July 9 launch, evidence the growth curve was still steepening rather than settling.
As of April 2026, Codex is included in Free, Go, Plus, Pro, Business, Edu, and Enterprise plans, with Free starting at $0/month.
Lower barriers to entry are also driving adoption of no-code test automation, letting teams without dedicated automation engineers build scalable testing workflows.
2. Usage & Adoption Statistics
Inside companies, OpenAI said Codex usage within ChatGPT Business and Enterprise grew 6x since January by April 2026, with 2 million+ builders using it weekly at that point.
Among ChatGPT Enterprise users, coding-related messages outside engineering, IT, and research rose 36% over six months. The most active users send 17x as many coding messages as the median worker.
Users who work across roughly 7 task types report 5x more time saved than those using about 4, a sign of a shift from isolated coding help toward multi-workflow agent use.
By June 2026, knowledge workers made up ~20% of Codex users, the fastest-growing segment, adopting it 3x faster than developers for research, spreadsheets, presentations, contracts, and data analysis.
Internally, OpenAI says that since August 2025 roughly 99% of its own Engineering work and 88-91% of Legal, Finance, and Recruiting work now runs through Codex.
OpenAI’s enterprise case studies back this up: Cisco used Codex to cut code review time by 50% and shrink project timelines from weeks to days.
3. Demographics Statistics
OpenAI hasn’t published Codex-specific demographic data, so the closest public proxies come from broader developer surveys.
Stack Overflow’s 2025 survey (33,662 respondents) found 84% using or planning to use AI tools, and 51% of professional developers using them daily.
That survey shows a clear experience gradient: 55.5% of early-career developers use AI tools daily versus 47.3% of experienced developers, and 53% of developers learning to code for AI used AI tools.
JetBrains’ 2025 survey of 24,534 developers found 85% regularly use AI tools for coding, development, and quality assurance, with 62% relying on at least one AI coding assistant, agent, or editor.
4. Revenue & Financial Statistics
OpenAI reported $20 billion+ in annual recurring revenue during 2025, with monthly revenue exceeding $2 billion by March 2026 and enterprise accounting for 40%+ of revenue.
For more, see our OpenAI Statistics breakdown.
Codex itself is priced at $0/month (Free), $8/month (Go), $20/month (Plus), and from $100/month (Pro). Business workspaces can use pay-as-you-go, token-based billing with no fixed monthly cost for Codex-only seats.
5. Enterprise Adoption Statistics
As of November 2025, OpenAI reported 1 million+ business customers worldwide, 7 million+ ChatGPT for Work seats, and ChatGPT Enterprise seats up 9x year over year, giving Codex a large built-in distribution base.
OpenAI’s April 2026 enterprise report also shows where coding tools sit in company budgets.
In finance, coding and developer tools ranked second among AI use cases; in technology, third, behind in-app assistants/search and agentic workflow automation.
6. Market Share & Competitive Statistics
GitHub Copilot remains the largest coding assistant by disclosed user count: 20 million+ users as of mid-2025, growing to 4.7 million paid subscribers by January 2026. (See our full GitHub Copilot statistics breakdown)
GitHub also reports 3 billion+ code suggestions accepted to date, 1.2 million pull requests a month, and use by 90% of the Fortune 100.
| Platform | Public User Figure | Other Scale Metric |
| Codex | 8M+ weekly (July 2026) | ~7-8x growth in 5 months |
| GitHub Copilot | 20M+ users, 4.7M paid | 1.2M PRs/month |
The competitive picture is shifting toward agents on both sides: Copilot is moving into agent mode and coding-agent workflows, while Codex is positioned as a cloud-based engineering agent that runs in parallel across toolchains.
The competition is now agentic workflow versus agentic workflow, not autocomplete versus autocomplete.
7. Developer, API & Platform Statistics
Codex sits inside a much larger OpenAI developer platform.
At DevDay 2025, OpenAI cited 4 million developers building on its platform, 800 million+ weekly ChatGPT users, API throughput of 6 billion tokens per minute, and API traffic up 5x year over year.
Codex’s last publicly disclosed benchmark set is from GPT-5.3-Codex (February 2026).
This ran 25% faster than its predecessor and scored 56.8% on SWE-Bench Pro (up from 56.4%), 77.3% on Terminal-Bench 2.0 (up from 64.0%), and 64.7% on OSWorld-Verified (up from 38.2%).
Codex now runs on newer models by default.
Sol for most users, Terra for Free/Go plans, and an Ultra mode with four parallel sub-agents for Pro and Enterprise, though OpenAI hasn’t yet published equivalent public benchmark scores for these.
As agents take on more complex workflows, organizations are also investing in AI automation testing to validate dynamic, agent-driven application behavior.
8. Codex CLI Statistics
Codex CLI, OpenAI’s open-source terminal agent launched in April 2025, has grown into one of the most popular AI coding tools on the market. By July 19, 2026, roughly 15 months after launch, it had passed 99,624 GitHub stars.
Its npm package downloads grew from 82,000 in April 2025 to 41.8 million by May 2026, a roughly 510x increase.
9. Regional & Geographic Statistics
OpenAI’s enterprise AI report shows where usage is rising fastest: Australia, Brazil, the Netherlands, and France posted 143%+ year-over-year growth in business customers.
The U.S., Germany, and Japan were the most active markets by business message volume.
The U.K. and Germany rank among the largest ChatGPT Enterprise markets outside the U.S. International API customer growth exceeded 70% over six months, and Japan is the largest corporate API market outside the U.S.
GitHub’s own developer map shows why this matters: India added 5.2 million+ developers in 2025, about 14% of GitHub’s 36 million new developers that year, bringing its total to 21.9 million on the platform.
10. Market Trends & Industry Growth
Grand View Research estimates the AI code assistants market at $8.51 billion in 2025, growing to $42.90 billion by 2033 at a 22.5% CAGR from 2026 to 2033.
OpenAI’s own usage data reflects a broader shift from AI-assisted coding to general-purpose AI work automation, evident in Codex’s move from developer-only milestones to a combined agent metric with ChatGPT Work.
As software delivery accelerates, engineering teams increasingly pair AI-assisted development with automated performance testing to confirm speed and stability before release.
Developer surveys support that trend: Stack Overflow found 84% using or planning to use AI tools, and JetBrains found 85% regularly using AI tools (like AI code review agents), with 62% relying on at least one assistant, agent, or AI editor.
Codex’s own messaging now spans feature writing, review, refactors, sandboxes, CLI use, IDE extensions, code review, and parallel tasks.
This is a sign the category is converging on end-to-end software delivery, not line-level suggestions.
Conclusion
Codex has kept expanding through 2026, growing from under 1 million weekly users in February to 7-8 million by mid-July, then merging with ChatGPT Work into a combined agent metric near 10 million.
Knowledge workers made up ~20% of users as of June 2026 and were adopting Codex 3x faster than developers, underscoring its shift from a coding assistant into a broader AI productivity tool.
Its last publicly benchmarked model, GPT-5.3-Codex, posted 56.8% on SWE-Bench Pro and 77.3% on Terminal-Bench 2.0, helping explain the pace of adoption even before GPT-5.6’s July 2026 launch accelerated growth further.
With OpenAI’s business base at 1 million+ customers, GitHub Copilot at 20 million+ users, and the broader AI code assistants market estimated at $8.51 billion in 2025, Codex competes in a category still expanding fast and moving toward agentic workflows.
FAQs
Q: How many users does Codex have in 2026?
A: Codex crossed 7 million weekly users on July 13, 2026, and reportedly reached 8 million within days. Since a July 9 product merge, OpenAI reports Codex alongside its new ChatGPT Work agent as a combined figure, which hit ~10 million by July 21, 2026.
Q: How fast is Codex growing?
A: Very fast: under 1 million weekly users in February 2026 to 7-8 million by mid-July, roughly a 7-8x increase in five months, accelerating further after the GPT-5.6 launch.
Q: What does Codex cost?
A: Free at $0/month, Go at $8/month, Plus at $20/month, and Pro from $100/month. Business workspaces can use pay-as-you-go, token-based billing instead.
Q: What are Codex’s best public performance stats?
A: GPT-5.3-Codex, the last model with published benchmarks, scored 56.8% on SWE-Bench Pro, 77.3% on Terminal-Bench 2.0, and 64.7% on OSWorld-Verified, running 25% faster than its predecessor.
Q: Who uses Codex the most?
A: Developers remain the largest group, but knowledge workers reached ~20% of users by June 2026 and were growing 3x faster. Coding-related messages outside engineering, IT, and research grew 36% over six months.
Q: How does Codex compare with GitHub Copilot?
A: Copilot still leads on disclosed user count, with 20 million+ users and 4.7 million paid subscribers, versus Codex’s 8 million weekly actives. Both are moving toward agentic coding workflows.
Q: Is there public data on Codex demographics?
A: OpenAI hasn’t published Codex-specific demographics. The closest proxy is broader developer surveys, where 51% of professional developers and 55.5% of early-career developers use AI tools daily.
