OpenAI Agents 10M Users: ChatGPT Work and Codex Fuel Record

Less than two weeks after launching ChatGPT Work, OpenAI has crossed the OpenAI agents 10 million users milestone. The total combines usage of Codex, its coding tool, and ChatGPT Work, an agent that automates multi-step tasks across apps and files. The figure, disclosed to Bloomberg ahead of a July 21 announcement, nearly doubles the roughly 6 million users reported when ChatGPT Work was first announced earlier this month, before the product had reached paying customers. Sam Altman said combined usage of the two agentic products jumped 2.5 times in a single week, with non-technical workers now the fastest-growing cohort. This milestone marks a turning point: users are moving from single-turn queries to agents that complete full workflows, signaling a shift in enterprise AI adoption.

The count measures weekly active users across both products rather than paid enterprise seats or independently audited metrics, but the growth trajectory is steep by any standard.

How ChatGPT Work Fueled the OpenAI Agents 10 Million Users Growth

ChatGPT Work launched on July 9. It automates multi-step tasks that span a user's applications and files. Instead of generating a block of text that the user then has to act on, the agent connects to services such as Slack and Google Drive to build spreadsheets, assemble slide decks, or move data between systems autonomously. The positioning puts it in direct competition with Anthropic's Cowork, and the two products are now the most visible entries in a rapidly growing category of AI agents aimed at knowledge workers.

The same day ChatGPT Work debuted, OpenAI folded Codex into the ChatGPT desktop app and released GPT-5.6 to the public. That integration gave Codex a wider distribution channel and likely contributed to the adoption spike that followed. Codex had already been scaling quickly. OpenAI reported more than 4 million weekly developers using it by late April 2026, up from roughly 600,000 at the start of the year. By June 2, the company put that number above 5 million weekly Codex users.

Non-Technical Users Reshape the Adoption Curve

The fastest growth is coming from workers who do not write code. OpenAI research showed that non-developer adoption of Codex grew 189 times compared to August 2025, outpacing the initial adoption rate among software engineers. By May 2026, more than 80 percent of sampled individual users had made at least one Codex request estimated to require extended, multi-step execution, a pattern that signals the tool is being used for substantive work rather than single-turn queries.

This shift matters strategically because it widens the addressable market for agentic AI far beyond the developer base that formed the early adopter pool. Knowledge workers in legal, recruiting, marketing, and finance departments now make up roughly 20 percent of Codex users, a segment that OpenAI says is growing three times faster than the developer cohort. If that trajectory holds, the majority of active users on OpenAI's agent platform could eventually be non-technical.

OpenAI published research characterizing this as one of the fastest documented shifts in professional tool usage. The dataset, which runs through June 11, 2026, shows that enterprise non-developer workers adopted Codex at a pace that dwarfed even the rate at which software engineers adopted the tool in its early months. Codex usage across teams grew six times since January and ten times since August 2025, according to the company's internal tracking.

Internal data illustrates how quickly the shift occurred within OpenAI itself. Through August 2025, the average OpenAI worker spent less than 10 percent of their tokens on Codex. By early 2026, every department including non-technical teams such as Legal and Recruiting had adopted Codex as their primary AI tool for work. The pattern suggests that the fastest path to enterprise-wide adoption runs through internal dogfooding, where employees demonstrate the value of the tool to colleagues in adjacent functions.

Enterprise Adoption and Early Use Cases

A set of early enterprise adopters illustrates the range of deployment scenarios. Nvidia, Virgin Atlantic, Zapier, and RingCentral are among the organizations using ChatGPT Work or Codex in production environments. For a company like Virgin Atlantic, an agent that can move between scheduling systems, passenger data, and operational tools in a single workflow is a tangible productivity gain. For Zapier, whose core business is automation, integrating an AI agent that can trigger and sequence its own workflows is a natural extension of its existing platform.

Alongside the user milestone, OpenAI launched a small business program aimed at bringing agentic tools to smaller organizations. The program extends the company's push beyond large enterprise accounts and into a segment that has historically been harder for premium AI platforms to reach because of cost and deployment complexity. Small businesses typically lack the in-house AI engineering teams that larger companies can deploy, so a product that finishes tasks across existing apps rather than requiring custom integration lowers the adoption barrier considerably.

The Competitive Picture

The rapid adoption of ChatGPT Work and Codex intensifies an already heated rivalry with Anthropic, whose Cowork product targets a similar use case. Both companies are racing to define the category of AI agents that act on behalf of knowledge workers inside the tools they already use. The stakes are high because the winner in this category could lock in enterprise workflows, creating switching costs that make it difficult for late entrants to dislodge an incumbent. OpenAI's lead in raw user numbers gives it a data advantage, as more usage means more training signals for improving agent reliability, but the category is young enough that positioning and execution still matter more than early market share.

OpenAI's revenue targets are ambitious, and the company is betting that agentic products will provide the growth engine needed to meet them. ChatGPT Work in particular moves the value proposition from a subscription for information retrieval to a subscription for task completion, which justifies higher price points and deeper integration into business operations. OpenAI has signaled internally that it expects the agent platform to become the primary revenue driver over the next several quarters, displacing the traditional chat subscription as the anchor product.

The broader market for agentic AI is still taking shape. Multiple startups and open-source projects are building competing approaches, but OpenAI's combination of an existing user base, a bundled product strategy that pairs coding and workplace agents under one platform, and aggressive distribution through the ChatGPT ecosystem gives it a structural head start. The question for enterprise buyers is no longer whether agentic tools are viable but which platform will deliver the most reliable results across the widest range of workflows. OpenAI's rapid user growth suggests it is winning the early race for attention, but the real contest will be decided by reliability, uptime, and the breadth of integrations each platform supports.

Why This Matters

Crossing the OpenAI agents 10 million users threshold in under two weeks after launching ChatGPT Work signals that demand for autonomous, multi-step agents is real and expanding beyond early adopters. For business decision-makers, the implication is that the category of AI tools worth evaluating has shifted from chatbots that answer questions to agents that execute work, and the window for building workflows around these systems is opening now, not in some distant future. The companies that figure out how to deploy agents effectively in the near term will hold a structural advantage as reliability improves and use cases multiply, while those that wait risk playing catch-up in a market that is already moving fast.

Sources

How agents are transforming work

Photo by Brett Wharton on Unsplash

✔Human Verified

Researched and cross-referenced against primary sources by the Bytevyte editorial team.