
OpenAI pairs its GPT-5.6 public rollout with ChatGPT Work, a new agent that handles entire workflows
Quick Answer
OpenAI has launched ChatGPT Work, powered by GPT-5.6, transforming its chatbot into an autonomous agent capable of managing entire workflows across various applications.
Quick Take
OpenAI has launched ChatGPT Work, powered by GPT-5.6, transforming its chatbot into an autonomous agent capable of managing entire workflows across various applications. This new product integrates Codex technology and features a Unified Plugins Directory for seamless third-party integrations, aiming to enhance productivity for Pro, Enterprise, and Edu users initially, with broader access to follow.
Key Points
- ChatGPT Work can generate outputs like Excel or Word documents from a single prompt.
- The Unified Plugins Directory includes integrations with major platforms like Google Drive and Slack.
- OpenAI claims 100% success in blocking data extraction attempts via the Auto-Review feature.
- Initial access to ChatGPT Work is for Pro, Enterprise, and Edu users, with wider rollout planned.
- Usage is based on task complexity and model selection, shared with Codex and other services.
📖 Reader Mode
~3 min readWith ChatGPT Work, OpenAI wants to turn its chatbot into an autonomous worker. The new product combines Codex technology with third-party integrations and runs on the just-released GPT-5.6.
About two weeks after the Trump administration restricted GPT-5.6 access to select organizations, OpenAI has now gotten approval for a public rollout. The company is also introducing a new product called ChatGPT Work for mobile, web, and desktop.
OpenAI describes ChatGPT Work as an agent inside ChatGPT that can operate across apps and files, work for hours on complex projects, and deliver finished outputs like an Excel or Word document based on a given task.
The product builds on Codex technology, which until now has mostly been reserved for developers. The new app is likely a step toward merging ChatGPT and Codex into a single "super app."
Plugins are back, and this time the models might keep up
The biggest addition in ChatGPT Work is a new "Unified Plugins Directory" that bundles third-party integrations in one place. At launch, the directory includes Google Drive, SharePoint, Slack, Microsoft Teams, Gmail, Outlook, Salesforce, Adobe, Zoom, LinkedIn, GitHub, Canva, and Dropbox, among others. Users can call on specific plugins with an "@" mention or let ChatGPT figure out which data source is relevant on its own.
The approach echoes OpenAI's failed plugin push from 2023. OpenAI co-founder Greg Brockman recently admitted that those plugins "didn't work at all because the models weren't ready." With GPT-5.6 behind it, OpenAI expects a different outcome this time.
According to OpenAI, ChatGPT Work can handle entire workflows from a single prompt. One example: turning customer research into a campaign brief, generating marketing assets from it, and adapting them for different markets, all while keeping context across every step.
From the announcement, it's hard to tell what's actually new. Many of the featured functions, like Scheduled Tasks, plugins, or Computer Use, already exist in ChatGPT or Codex, and ChatGPT could already handle long tasks and connect data sources before. Similar to Anthropic's shift from Claude Code to Cowork, this looks moslty like a rebrand of Codex for knowledge work with local file access, as long as users run the desktop app. Anthropic also just launched Cowork for web and mobile, likely a preemptive move because they knew ChatGPT Work was coming.
On security, OpenAI points to an "Auto-Review" feature where advanced models check important actions before they run. OpenAI says Auto-Review blocked 100 percent of attempts to extract protected data during adversarial red-teaming.
Phased rollout, usage-based billing
On web and mobile, Pro, Enterprise, and Edu users get access to ChatGPT Work first. OpenAI says Plus and Business customers will follow in the coming days. Through the ChatGPT desktop app for Mac and Windows, though, the product is available right away for all plans, including the free tier. As of this writing, the Windows link on the download page still just leads to the Codex app.
Usage works differently from standard chat. According to OpenAI, ChatGPT Work shares the same agent-based consumption pool with Codex, ChatGPT for Excel, and Workspace Agents. How much a task consumes depends on its size, complexity, and the model selected. Enterprise and Edu admins can use Spend Controls to manage limits at the workspace, group, and individual user levels.
The Codex pricing page lists included quotas and extra credits, though OpenAI notes that the examples there are based on coding tasks and consumption for Work tasks may differ. Once the quota runs out, some Codex Plus and Pro users can buy more credits, while others have to upgrade or wait for it to reset.
— Originally published at the-decoder.com
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from The Decoder
See more →
An AI model programmed nonstop for 19 days on a single MirrorCode task that cost $2,600 to run
Epoch AI's MirrorCode benchmark reveals Claude Opus 4.7 as the leader with a 56% solve rate, reconstructing a 16,000-line toolkit in 14 hours. Despite this, all models tested struggle with the most complex tasks, highlighting limitations in current AI capabilities. The single task consumed $2,600 over 19 days, raising questions about cost-effectiveness in AI development.

