OpenAI launches Dots free on Pro, then cuts Pro’s allowance to 10x

OpenAI handed every Pro subscriber a free dot at DevDay on Tuesday, and in a separate announcement the same day it cut back the allowance those dots run on. From 30 October the $200 Pro plan’s included usage in Codex and ChatGPT Work drops from 20 times the Plus allowance to 10 times. The two announcements ran as separate stories, so almost nobody set them side by side.

Dots are the always-on agents OpenAI announced in San Francisco, powered by GPT-6 Astra. CNBC reported that each dot has its own cloud computer and connects to more than 4,000 apps. You can message one inside ChatGPT, Slack and Teams, and texting is promised later. Sam Altman called them “remarkably capable, always-on agents built to handle really anything you can think of” during the keynote.

Availability is narrower than some write-ups suggested. CNBC reported dots roll out in ChatGPT for Pro and Business Premium users in eligible markets, while Enterprise, Edu and Healthcare workspaces get them only once an administrator enables them. Engadget’s launch piece listed Enterprise alongside Pro as having access at launch, which isn’t what CNBC’s account says.

What a free dot actually costs

The pricing news ran as its own story. OpenAI reopened the $200 Pro plan to new subscribers and added Pro 500 at $500 a month, and The New Stack reported that the new tier carries 25 times the Plus allowance. That same report dates the cut to the older plan at 30 October, so the two moves sit a month apart.

Plan	Price	Codex and Work allowance vs Plus	GPT-6 Pro messages a week
Pro, through 29 October	$200 a month	20x	200
Pro, from 30 October	$200 a month	10x	100
Pro 500	$500 a month	25x	not disclosed

That matters for dots because of how OpenAI meters them. Engadget reported that conversations with your dot don’t count against your ChatGPT usage limit, and its launch piece says usage limits in Codex and Work apply as normal. So the free dot is free to talk to rather than free to run, and the budget it draws down is the one being halved.

OpenAI’s head of product and platform, Tibo Sottiaux, previewed the change on X ahead of the official announcement and argued the arithmetic still favours subscribers. The New Stack quoted his post directly.

In effect, if you do the math, it will net out at half the dollar in API spend compared to the old Pro $200 plan.

Tibo Sottiaux, head of product and platform, OpenAI, via The New Stack

Existing Pro subscribers keep their current limits through 29 October. After that they get 62,500 usage credits, which OpenAI values at $2,500, expiring 31 December 2026. Sottiaux’s case is that cheaper models close the gap, and The New Stack’s Frederic Lardinois wrote that efficiency gains aren’t what people bought the plan for. That report also notes the 20x allowance is still the standard at Google’s top Gemini AI Ultra plan and Anthropic’s top tier, both of which cost $200 a month.

The safety framing got less attention than the avatars, though it’s the part that bounds the product. On Hacker News, where the thread passed 640 points and 500 comments inside a day, one commenter quoted OpenAI’s own wording. The background “proactive research” mode uses connected apps with tools restricted to read-only. In OpenAI’s words, that means they can’t send messages, change app content, or control your browser or computer.

That’s a real limit on what “always-on” means, and it lands in the middle of the agent incidents CNBC says have put OpenAI under intense pressure.

The domain joke the record doesn’t support

Meanwhile, most of the follow-up coverage went elsewhere. TechCrunch and Business Insider both reported that dot.com redirects to a Grok download page, read by many as a prank on the launch.

We checked it ourselves on Wednesday. The address returns a 301 to x.ai/bot, and the registry record was last updated on 28 July 2026. The registrant is listed as Perfect Privacy, LLC, a Network Solutions proxy.

So the public record supports the July timing that both outlets cited, but it names no owner at all.

Chief financial officer Sarah Friar told CNBC in a live interview that the premium gating is temporary. She said OpenAI wants dots in front of its 1.2 billion consumer users, as Business Insider reported. Altman said the same in a DevDay question session, describing the premium start as deliberate because it helps people understand how expensive the work is. Neither gave a date or a price for the wider rollout.

What to watch is whether the 30 October cut survives the developer pushback The New Stack recorded, because it sets the real ceiling on what a free dot can do. OpenAI has been cutting prices on its cheaper models at the same time, so the squeeze is specific to the tier the agents launched on.