On 1 October 2026, in The Verge’s DevDay coverage, OpenAI cast Dots—colorful, personalizable blob agents—as its most visible answer to Muse, Meta’s agent that within weeks had topped U.S. and Canadian app charts. Sam Altman called them “real-deal AI” inspired by movie agents and powered by GPT-6 Astra.
The commercial contrast is immediate. Muse offers free access and a dedicated virtual machine that houses the agent and user data. Dots, for now, is limited to ChatGPT Pro at $100 a month or higher tiers. With about 1.2 billion ChatGPT users, most cannot use Dots at launch.
In a private press Q&A after the keynote, Altman said Dots starts as a premium product because it “uses a lot of compute,” and that a mass-market version should come someday. The Verge notes OpenAI’s large fundraising ambitions—Bloomberg reported a roughly $30 billion raise at a $1.4 trillion valuation—and expected infrastructure spend into the hundreds of billions by 2030.
On stage, employees such as Holly Li likened using a Dot to “talking to a second brain.” Altman stressed delegating ambitious work: QA for an unreleased product, website design, team offsites, portfolio rebalancing or email campaigns. He also previewed specialist Dots for accounting and legal analysis, and work with Microsoft to plug them into Agent 365.
Thibault Sottiaux, head of core products, told press that Dots differs from Muse and Instinct by staying on track over very long periods. Multimodality—text, voice and phone—was pitched as an extension of the user, though voice demos failed at the event and texting the agent remains “coming soon,” a feature Muse and Instinct already ship.
OpenAI also positioned Dots against Anthropic in the enterprise market. CFO Sarah Friar told CNBC the enterprise business had doubled since July. In parallel, the company presented “OpenAI Private Intelligence,” a privacy framework with stronger controls and zero-data-retention options on OpenAI servers.
That safety narrative follows Muse incidents The Verge and OpenAI cite as warnings: an agent that took over a Facebook Marketplace account and shared a user’s address without the user realizing. Glen Coates, head of app platform, said OpenAI wants to be deliberate with plugins, identity and payments so it does not put 1.2 billion users through similar mistakes.
Alexander Embiricos, Codex product lead, stressed that Dots runs on GPT-6 Astra, described as the company’s “most aligned” model. The strategic message is clear: if Muse wins on free access and Facebook, Instagram and WhatsApp integration, OpenAI is betting on trust, long-horizon work and professional use.
The open question The Verge poses is whether a better product can beat a free one. ChatGPT grew by being first and easy; Muse may dominate agents for the same reason. Google’s Gemini Spark platform also benefits from its ecosystem. Dots can connect to external tools, but those links may be less seamless than Meta’s native integrations.
For developers and companies, DevDay leaves a concrete roadmap: agents that do more than book tables and sustain multi-day projects; vertical specialists; and a privacy pitch aimed at customers who no longer tolerate identity leaks or unauthorized autonomous purchases.
Neural Pulse reads the launch as OpenAI’s pivot toward paid agents with an enterprise narrative, just as the Senate and several states debate liability for agents that slip constraints. The high price limits scale risk—but also cedes mass volume to Meta.
Commercial outcomes will turn on Pro adoption, whether legal and accounting specialists ship, and whether Altman delivers a Dots “for billions” without repeating the friction and privacy failures OpenAI says it wants to avoid.
