According to official OpenAI announcements and reports from 9to5Mac and Decrypt, OpenAI introduced Workspace Agents—cloud agents powered by the Codex model that teams can build together and share—on April 22 in the ChatGPT Business, Enterprise, Edu, and Teachers plans. They can continue executing long-running workflows even when users are offline. The research preview period is completely free until May 6; afterward, it will use a credit-based billing model.
Codex-powered: cloud long-running agents, cross-tool context preserved
Workspace Agents run on OpenAI’s Codex model and continue the capabilities introduced by the April 16 major update, “Codex for (almost) everything”: computer use, in-app browsers, gpt-image-1.5 image generation, memory preview, 90+ plugins, and parallel agents.
The difference is that Workspace Agents are designed for team sharing, cloud operation, and long-lasting continuity: users set it up once, and the entire team can call on it. When the agent runs, it doesn’t require humans to be online; it can handle tasks ranging from preparing reports, writing code, and replying to messages to multi-step workflows. OpenAI’s own product team example is that they built an agent that “proactively answers employees’ questions in Slack channels and automatically opens a ticket when new topics are discovered.”
Native Slack integration: agents operate like team members
Workspace Agents can be deployed inside ChatGPT or to Slack channels: on the Slack side, agents can proactively pull messages in channels, respond to questions, execute workflows, and open issues/tickets. This is OpenAI’s first time integrating ChatGPT agent capabilities to a depth that’s “like a Slack bot, but stronger.”
Scheduling is also in place: agents can be configured with daily/weekly/custom cron triggers, with no manual intervention needed. This means small teams can use Workspace Agents to replace some of their existing automation scripting tools (such as Zapier and n8n).
Plans and billing: free during the research preview until 5/6, then switch to credit
Available plans and billing:
ChatGPT Business
ChatGPT Enterprise
ChatGPT Edu (for education)
ChatGPT Teachers (for teachers)
Before May 6, 2026, Workspace Agents are free for the research preview phase. After that, it will switch to credit-based billing—each time an agent executes, it consumes credits; the exact rates have not been announced yet. In the past, OpenAI’s credit pricing for Codex and Operator was about $5–10 per thousand credits, which can serve as a reference.
Three major enterprise players’ agents clash in the same week
The launch timing for Workspace Agents coincides with Google Cloud Next 2026 unveiling the Gemini Enterprise Agent Platform, and Anthropic releasing the Claude Cowork Live Artifacts in the same week. Over the same week, the three AI leaders each unveiled their enterprise agent strategies:
OpenAI Workspace Agents: Codex-powered, native Slack, fast deployment by individual teams
Google Gemini Enterprise Agent Platform: full-stack integration, $750 million in subsidies for consulting firms
Anthropic Claude Cowork + MCP: Live Artifacts real-time dashboards, and an open MCP ecosystem
In their respective visions of how AI agents enter enterprise workflows, each of the three has its own emphasis, and this year’s second half of enterprise AI procurement will enter a first-ever “three-way head-to-head showdown” phase among the top three.
This article about OpenAI launching ChatGPT Workspace Agents: Codex-powered, team sharing, Slack integration first appeared on ChainNews ABMedia.
Related Articles
Delphi AI Prediction Market Launches on Gensyn Mainnet
MetYa and Zypher Network Partner to Build Privacy-First Web3 Platform with AI and Zero-Knowledge Tech
OristaPay Launches Telegram AI Payment Gateway with TON and BytePlus
GoPlus AgentGuard Launches Checkup Mode to Strengthen AI Agent Web3 Security
Anthropic Launches /ultrareview for Claude Code: Multi-Agent Cloud-Based Code Review
OpenAI Launches ChatGPT Workspace Agents for Enterprise Workflow Automation