OpenAI's Agent User Base Grows to 10 Million, Driven by Enterprise Workflow Adoption

Enterprise monetization signals are materializing: Microsoft 365 Copilot has reached 15 million paid seats, about 160,000 organizations have deployed over 400,000 Copilot Studio agents, and Salesforce Agentforce closed 29,000 deals at roughly $800 million ARR, indicating real spend embedding agents into workflows.
OpenAI’s commercial model is shifting away from a guaranteed revenue flow via Microsoft: OpenAI products will ship first on Azure, but the company can distribute across clouds, reducing Microsoft’s exclusivity moat and altering the economics of the partnership.
The 10 million-user figure is a combined tally for Codex and ChatGPT Work due to a single redesigned ChatGPT app and shared usage pool, making it difficult to split users cleanly by product or define what 'active' means.
Codex’s footprint was already meaningful before the 10M milestone: the platform logged more than 5 million weekly Codex users and over 1 million non-development users leveraging the platform for tasks beyond coding.
GPT-5.6’s general availability and the Sol tier accelerated adoption by targeting coding, knowledge work, and long-running tasks, helping demand climb rapidly once paying users were available.
OpenAI's agent tools have crossed 10 million weekly active users, nearly doubling in less than two weeks, according to The Next Web. The milestone covers both Codex, its coding agent, and ChatGPT Work, a system that handles multi-step tasks across apps and files after launching on July 9.
Codex lead Tibo Sottiaux confirmed the figure, Digg reported. Before hitting 10 million, Codex alone had logged more than 5 million weekly users, with over 1 million of those using it for tasks beyond coding.
The 10 million figure combines both Codex and ChatGPT Work into a single number. That's because OpenAI merged them into one redesigned ChatGPT app with a shared usage pool, TechBooky reported. This makes it hard to know how many users each product has on its own.
The definition of "active" is also unclear. OpenAI has not said what counts as an active user or how often users must return to qualify. That gap matters when investors and partners try to judge real demand versus one-time sign-ups.
Much of the growth came after OpenAI made GPT-5.6 broadly available. The new model targets coding, knowledge work, and long-running tasks. A new tier called Sol opened the door for paying users, and demand climbed fast once that option was live, The Next Web reported.
OpenAI also launched a ChatGPT for small business program at the same time, 9to5Mac reported. The program gives small businesses access to the same agent tools, aiming to push adoption beyond large enterprises and into everyday workflows.
The broader market is already spending real money on agent tools. Microsoft 365 Copilot has reached 15 million paid seats. About 160,000 organizations have deployed over 400,000 Copilot Studio agents. Salesforce's Agentforce closed 29,000 deals at roughly $800 million in annual recurring revenue.
These numbers show that agents are moving from demos into daily business use. The key question for OpenAI is how many of its 10 million users become repeat, paying customers. Even a small conversion rate at scale would mean significant new revenue.
The growth story matters differently for Microsoft now. Since April, OpenAI changed its Azure deal. OpenAI products still ship on Azure first, but the company can now sell across other clouds too. That breaks Microsoft's exclusivity advantage, Healthcare Finance News reported.
Microsoft remains a major cloud partner and a key platform for OpenAI tools. But OpenAI growth no longer flows automatically into Microsoft revenue. Investors watching both companies need to track each one's monetization separately going forward.
Publishers
10
Articles
5
Reach
15