OpenAI launches Dots agent on GPT-6 Astra, limited to $20/month Pro tier and above

OpenAI has launched Dots, a personal AI agent powered by GPT-6 Astra, but access is restricted for now to subscribers on its $20/month Pro plan and above. Meta's competing Muse agent platform is free. OpenAI announced Dots at its DevDay 2026 conference.

What OpenAI announced

CEO Sam Altman introduced Dots as a "real-deal AI" agent. Each agent is called a "Dot" and takes the form of a customizable, colorful blob with eyes, a design that closely mirrors Muse's approachable styling.

According to OpenAI, Dots is built for delegating long-running, "ambitious pieces of work," including:

• QA for an unreleased product
• Website design
• Team offsite planning
• Portfolio rebalancing
• Email campaign creation

OpenAI also demonstrated Dots building virtual worlds. Thibault Sottiaux, OpenAI's head of core products, said Dots is "capable of actually staying on track in very long periods of time." Altman claimed the product is "far beyond anything else available on the market today." No independent benchmarks or evaluations were published.

Availability and pricing

• Access: Pro plan ($20/month) and up. Most of ChatGPT's 1.2 billion users cannot use Dots yet.
• Compute cost: Altman said Dots "uses a lot of compute" and is "starting out as a premium product." He said a mass-market version for "billions of people" should be expected eventually, with no timeline given.
• API pricing, context window and benchmark scores: not yet disclosed.

Features still pending

OpenAI said "specialist" Dots for areas such as accounting and legal analysis will launch soon. The company is working with Microsoft to integrate Dots into its Agent 365 platform.

Dots is also multimodal, with text, voice and phone-call interaction. The Verge reports that voice demos failed during DevDay. Texting an agent, which OpenAI highlighted repeatedly, is still listed as coming soon. Muse and another agent, Instinct, already support it.

Privacy and safety positioning

OpenAI describes GPT-6 Astra as its "most aligned model." At DevDay it also introduced "OpenAI Private Intelligence," a data privacy framework that gives businesses "stronger controls" over their data, including zero data retention options. Altman said OpenAI wants to "set a new standard for privacy in frontier AI."

The pitch is aimed at Meta. Muse was recently criticized after it reportedly took over a user's Facebook Marketplace account and gave a stranger the user's address without the user's knowledge. Glen Coates, OpenAI's head of app platform, said OpenAI is being "really deliberate" about how agents use plug-ins, identities, sign-ins and payments.

Financial context

The pricing decision lands against a heavy cost backdrop, per outside reports:

• OpenAI is reportedly seeking $30 billion at a $1.4 trillion valuation (Bloomberg).
• Axios reports annual recurring revenue is nearing $70 billion after 70% growth since the start of Q3.
• The FT reports OpenAI expects to spend $280 billion by 2030.

OpenAI CFO Sarah Friar told CNBC the company's enterprise business has doubled since July.

What this means

OpenAI is not trying to win the consumer agent race on price. Meta can give away a hosted virtual machine per user and plug Muse into Facebook, Instagram and WhatsApp. Google's Gemini Spark has the same ecosystem advantage. OpenAI's connectors to external tools will likely be less seamless than those first-party integrations.

Dots is better read as an enterprise and prosumer play. Long-horizon task reliability, specialist agents, Agent 365 integration and zero-retention controls are aimed at the market where Anthropic has led. Those are also the claims that most need independent verification, and none has been tested outside OpenAI's own demos. The failed voice demos and the unreleased texting feature suggest the multimodal story is not yet ready.

The open question is whether a paywalled, compute-heavy agent can build the usage base needed to justify its cost before a free competitor locks in consumers. Altman's promise of a mass-market version signals that OpenAI expects to cut costs, but it has given no date.