OpenAI has launched ChatGPT Work, a new agentic feature embedded in ChatGPT designed to turn the chatbot from a question‑and‑answer tool into an autonomous workplace platform. The product runs on OpenAI's latest flagship model, GPT‑5.6.
Capabilities
ChatGPT Work is intended to do more than generate text: it gathers context from connected apps, files and workflows to produce finished outputs such as documents, spreadsheets, presentations, reports and websites. The agent accepts a stated outcome, breaks it into sub‑tasks, and can continue working on complex projects for hours to complete them independently.
Its core architectural bet is a persistent cloud virtual machine running on OpenAI's servers, available across devices. OpenAI positions this as a deliberate contrast to competitors whose agents require a local machine to stay powered on.
Availability and pricing tiers
The rollout begins with Pro, Enterprise and Edu users, expanding to Plus and Business users over the following days. OpenAI emphasized that making ChatGPT Work available to Plus subscribers — not only premium enterprise tiers — is central to its strategy; Plus costs $20 per month.
Connectivity and extensions
ChatGPT Work connects to external services via MCP‑based plugins, including Gmail, Google Calendar, Slack and GitHub. OpenAI confirmed these plugins are built on the Model Context Protocol (MCP) standard, and support for connecting multiple Gmail accounts is on the roadmap.
The user experience is action‑oriented from the outset, with a personalized onboarding flow that surfaces suggested use cases based on the user's role. For example, when the system detects a product manager role, it may suggest tasks such as reviewing the calendar, summarizing Slack activity, or evaluating AI systems.
Examples of agentic productivity
Ty Geri, a product manager who helped build ChatGPT Work, demonstrated real use cases. Ahead of launch he asked the system to schedule "bug bash" testing sessions across dozens of features and contributors; ChatGPT Work checked Slack, GitHub and Docs to identify the top contributors and scheduled ten bug bashes, a job that Geri says would have taken at least 30 minutes manually.
Geri also described using the tool for analytical work, such as diagnosing the main causes of user churn for specific features and generating product solutions — tasks he said would previously have taken months but can now be condensed to a week.
Data privacy and enterprise security
Because ChatGPT Work actively reads data from connected workplace tools — Slack messages, calendar invites, GitHub commit histories — privacy and control are prominent concerns. Ty Geri stated that user control is "incredibly important," and pointed to OpenAI's existing enterprise security infrastructure. He noted enterprise accounts have zoning and data retention (ZDR) features and that users can opt out of allowing their conversations to be used to improve future models.
OpenAI previously said at the ChatGPT Enterprise launch in August 2023 that it does not train models on customer business data or conversations. Nevertheless, ChatGPT Work's active context‑gathering represents a different data surface area than a user pasting text into a chat prompt, and enterprise security teams will likely scrutinize access carefully.
Competitive landscape
ChatGPT Work competes directly with Anthropic's Claude Cowork and Microsoft Copilot Cowork, both of which offer cloud‑based, persistent agents on web and mobile that break tasks into steps, connect via plugins to workplace tools, and produce finished outputs rather than only conversational replies.
OpenAI's consumer reach is a key differentiator: the company says ChatGPT has 900 million weekly active users and 50 million paying subscribers. More than 9 million paying business users rely on ChatGPT for work, and 92% of Fortune 500 companies use the service. By making ChatGPT Work available to Plus subscribers at $20 per month, OpenAI is betting that broad accessibility will accelerate adoption.
Financial context and IPO timing
The launch comes as OpenAI prepares for a potential IPO: the company confidentially filed a draft S‑1 registration statement with the SEC, and press reports place its valuation between $730 billion and $852 billion. OpenAI reports approximately $2 billion in monthly revenue, with enterprise accounting for more than 40% of revenue and expected to reach parity with consumer revenue by the end of 2026. The company also expects to remain loss‑making, not forecasting profitability until around 2030, and projects a $14 billion loss in 2026.
Anthropic filed for its own IPO on June 1 with a $965 billion valuation, setting up near‑simultaneous public offerings from two leading AI startups. Public investors will likely evaluate whether enterprise products like ChatGPT Work can deliver measurable productivity gains that convert into sustained revenue.
Product lineage and models
ChatGPT Work consolidates OpenAI's earlier enterprise efforts: the ChatGPT Enterprise launch (August 2023), the Operator agent (January 2025) and Operator's deprecation on August 31, 2025, when those capabilities were folded into the ChatGPT agent framework. ChatGPT Work pairs GPT‑5.6's three model variants — Sol (power), Luna (speed) and Terra (balanced everyday use) — with a persistent cloud environment and an expanding library of MCP plugins.
Implications for the future of work
Ty Geri declined to make sweeping predictions but suggested the product points toward a future where users interact with computing environments primarily through an AI agent. He described the agent as an "extension" and a partner that can interact with websites and a user's machine to perform tasks on their behalf.
Geri concluded with a personal remark that he has "never enjoyed work as much" as in the month using ChatGPT Work and Codex, a claim OpenAI now asks its 900 million users to scale. For a company preparing for one of the largest possible public offerings, whether that user experience scales is a question with major financial implications.
