OpenAI launches always-on Dots agents and a new $500 tier at DevDay

By Splitfeed EditorialLast updated Sep 29, 2026, 17:51 UTC

On September 29, 2026, at its annual DevDay conference in San Francisco, OpenAI unveiled Dots, a set of always-on AI agents built on a model called GPT-6 Astra. Dots run on their own cloud computers, can browse the web, fix bugs, send invoices, and connect to platforms including ChatGPT, Slack, and Microsoft Teams. When not actively prompted, they operate in the background with read-only access, scanning for ways to assist users. Alongside Dots, OpenAI announced a new Pro 500 subscription tier priced at $500 per month, a revamped ChatGPT with shared workspaces, collaborative documents and slides, an open plugin system with MCP event automation, and an enterprise marketplace featuring 32 partners. CEO Sam Altman opened the keynote by saying the company had found a "new thing" and promised more than 20 product launches at the event.

The announcement lands at a moment when the agentic AI space has become intensely competitive. Before DevDay, reporting noted that OpenAI had fallen behind rivals in the category of continuously running, consumer-facing agents. Dots are positioned as a direct answer to Meta's Muse agents, which are themselves facing resistance: NBC News reports that Amazon and Walmart are blocking Muse from operating in their shopping ecosystems. The $500 monthly tier represents a significant jump in pricing that raises questions about who the intended customer is and whether the productivity gains justify the cost. The open plugin system and enterprise marketplace suggest OpenAI is also trying to lock in business customers through an ecosystem strategy, not just a single product.

Because no published reactions from the Pro-AI, Anti-AI, or Middle Ground camps have surfaced yet, expectations have to stand in for stated positions. Supporters of AI development would typically highlight Dots as evidence that autonomous agents are maturing into practical tools, pointing to the concrete integrations and background automation as proof of real-world utility. Critics would likely raise concerns about always-on agents operating with persistent access to user data and third-party platforms, and would question whether the $500 tier prices meaningful AI assistance out of reach for most individuals. Those in the middle would probably focus on the unresolved tension between capability and control, asking what safeguards govern background read access and how the enterprise marketplace handles accountability when an agent makes a consequential mistake.

The arguments worth tracking in the weeks ahead include how developers respond to the open plugin system, whether the 32 enterprise marketplace partners report meaningful adoption, and how the competition with Meta's Muse unfolds, particularly if more retail platforms follow Amazon and Walmart in restricting third-party agents from their ecosystems.