OpenAI confirmed on June 2 that Codex has surpassed 5 million weekly active users, a sixfold increase since the desktop app launched in February 2026. The tool, originally designed as a coding assistant, is now being adopted by knowledge workers who use it for reports, spreadsheets, presentations, contracts, and workflow automation.
The growth is not just about raw numbers. Knowledge workers now represent roughly 20% of Codex’s user base and are growing more than three times faster than the developer segment. This shift is detailed in OpenAI’s report “The Next Era of Knowledge Work,” released alongside the user statistics.
What Knowledge Workers Are Doing With Codex
According to the report, the fastest-growing tasks among non-developer users are data analysis, research, and what OpenAI calls “knowledge artifact creation”. In practice, this means people in marketing, finance, operations, and HR are using Codex to pull data from multiple sources, generate formatted reports, and build lightweight tools that previously required engineering support.
Users are also running multiple Codex tasks in parallel. A project manager might simultaneously investigate sales data, draft a client presentation, and automate a reporting workflow, all through separate Codex sessions. This kind of parallel task execution is relatively new for AI assistants and points toward a different relationship between workers and AI tools.
The Growth Trajectory
The numbers tell a clear story of accelerating adoption:
- January 2026: approximately 600,000 weekly active users
- April 2026: 4 million weekly active users
- June 2026: 5 million weekly active users
The jump from 4 million to 5 million happened in roughly six weeks, even as the base got larger. OpenAI attributes part of this growth to the knowledge worker segment, where word-of-mouth and internal tooling adoption within companies is driving usage without direct marketing.
How Codex Differs From ChatGPT
While ChatGPT handles conversational queries, Codex operates differently. It works within a sandboxed environment, executes code, accesses files, and produces structured output. For a knowledge worker, this means Codex can actually build a spreadsheet from raw data, format a presentation deck, or create a PDF report rather than just describing how to do those things.
The distinction matters for enterprise adoption. ChatGPT is a general-purpose chatbot; Codex is a task execution engine that happens to use natural language as its interface.
Industry Impact and Concerns
The data suggests a structural shift in how non-technical workers interact with computing tools. Previously, a marketing analyst who needed a custom report would file a request with the IT department and wait days or weeks. Now they describe what they need to Codex and get a working result in minutes.
This creates friction within organizations. IT departments worry about shadow AI, where employees use Codex to process sensitive data outside approved channels. Security teams are also concerned about the implications of giving non-technical staff the ability to write and execute code, even in a sandboxed environment.
OpenAI’s report acknowledges these concerns but argues that the productivity gains outweigh the risks, particularly when organizations implement appropriate guardrails around data access and output sharing.
What’s Next for Codex
OpenAI has not announced specific feature updates tied to the knowledge worker segment, but the growth data suggests the company will invest more in enterprise-oriented capabilities. The parallel task execution feature, which allows users to run multiple Codex instances simultaneously, is likely to see further development as it becomes a key differentiator for business users.
At 5 million weekly users, Codex has become one of the fastest-growing AI products in the market. The question now is whether it can maintain this trajectory as competitors like Anthropic’s Claude and Google’s Gemini add similar code execution and task automation features.
FAQ
What is OpenAI Codex used for by non-developers?
Knowledge workers use Codex to create reports, spreadsheets, presentations, and contracts. They also use it for data analysis, research, and automating workflows that previously required engineering support.
How many users does OpenAI Codex have?
As of June 2026, Codex has over 5 million weekly active users, up from approximately 600,000 at the start of the year.
Is Codex different from ChatGPT?
Yes. Codex operates in a sandboxed environment that executes code, accesses files, and produces structured output. ChatGPT is conversational; Codex is a task execution engine that uses natural language as its interface.
Can Codex replace coding assistants like GitHub Copilot?
Codex serves a broader audience than Copilot. While Copilot focuses on inline code suggestions for developers, Codex handles entire tasks end-to-end, making it accessible to non-technical users who want to automate work without writing code themselves.
What security concerns exist with enterprise Codex usage?
IT departments worry about shadow AI (employees processing sensitive data outside approved channels) and the implications of giving non-technical staff code execution capabilities, even in sandboxed environments.
