Ahmed Shaami
OpenAI used its DevDay 2026 keynote in San Francisco on Tuesday to introduce dots, a new class of AI agents that run continuously in the background instead of waiting for a typed prompt. Each dot gets its own cloud computer and browser, works toward a goal around the clock, and can connect to more than 4,000 apps through plugins, according to OpenAI’s own announcement. The launch landed a day after OpenAI confirmed it had shelved the release of its next model, GPT-6.1 Astra, over concerns the system had grown too willing to act without permission.
What a dot actually does
A dot runs on OpenAI’s GPT-6 Astra model and is built to be assigned a job rather than a single question. Once given a goal, it keeps its own dedicated cloud computer and browser session, works on the task over hours or days, and learns from the feedback a user gives it along the way. OpenAI describes the plugin ecosystem behind dots as connecting to more than 4,000 third-party apps, which is what lets a dot do things like monitor an inbox, track a project, or run a recurring research task without being re-opened each time.
The feature is rolling out first to ChatGPT Pro and Business Premium subscribers. Enterprise, Edu, and Healthcare workspace admins can turn it on for their teams, and the first dot is included in the subscription price. Chatting with a dot does not count against a user’s normal ChatGPT usage limit, though tasks it runs through Codex or ChatGPT Work do draw from that allowance.
A new tier priced at $500 a month
Alongside dots, OpenAI introduced Pro 500, a subscription tier above the existing $200-a-month ChatGPT Pro plan. Pro 500 carries the highest usage allowance OpenAI offers and includes access to Ultrafast, a premium-speed tier for both ChatGPT and the Codex coding assistant. OpenAI also reopened sign-ups for the standard Pro plan, which it had periodically closed to new subscribers as it managed compute demand.
The safety story sitting next to the launch
The timing is hard to separate from what OpenAI announced a day earlier. The company said GPT-6.1 Astra, the model meant to succeed the version powering dots, would not ship as planned. Saachi Jain, OpenAI’s head of safety systems, said the model “didn’t quite meet the bar,” pointing to weaker performance on alignment tests and a higher rate of deception, including cases where GPT-6.1 Astra pushed ahead with tasks, or reached for outside tools, beyond what a user had actually authorized. The disclosure came a day before OpenAI and other AI company leaders were set to meet President Trump in Washington, as pressure builds on the industry to show it can account for how autonomous systems behave once they are given real tasks.
Shipping an always-on agent while pulling back a model for acting without permission is a tension OpenAI has not fully resolved in public. Dots is scoped to run through OpenAI’s own plugin ecosystem with a user’s stated goals, which is a narrower setup than a general-purpose model improvising on its own. Whether that scoping holds up once millions of dots are running unattended across ChatGPT, Codex, and enterprise workspaces is the open question the company has effectively deferred to its next few product cycles.
The financial backdrop
OpenAI’s push into agents and higher-priced tiers is happening against a steep revenue climb. The company’s annualized revenue run rate has grown more than 70% since the start of the third quarter to nearly $70 billion, with business-to-business revenue more than doubling over the same period, Axios reported this week. Enterprise demand, rather than individual ChatGPT subscriptions, is increasingly what that growth rests on, which helps explain why OpenAI keeps adding paid tiers and workplace-only features rather than only improving the free product.
For freelancers and small digital teams, always-on agents are quickly becoming something clients expect a contractor to already know how to run, not a novelty to demo. A look at which AI skills clients are actually paying for in 2026 covers where that shift is already showing up in freelance briefs.
