Three Weeks In: What OpenAI Still Hasnt Told Dots Users
Dots launched three weeks ago with a grace period, a narrow rollout, and several open questions. Most of them are still open — here's the running list.
Three weeks after dots launched on September 29, 2026, it's worth separating what OpenAI has actually said from what the coverage has assumed on its behalf. The honest answer, checked against OpenAI's own release notes and the best third-party reporting available as of October 22, is that most of the open questions from launch week are still open.
Why a running list, rather than a single verdict
Most coverage of a product a month into its life tends to collapse into a single verdict — "dots is succeeding" or "dots is struggling." Both framings require more certainty than the available evidence supports. A running list, checked against OpenAI's own statements rather than inferred from launch-week momentum, is a more honest format for a product whose most consequential open questions are about things OpenAI hasn't said yet, not things it has said and been caught contradicting. That's a genuinely different, and better, starting point than a headline grade would be at this stage.
What's confirmed, and from where
OpenAI's own changelog records a real string of activity in October: Locked Chats rolling out to Free, Go, Plus, and Pro users across web, iOS, and Android on October 20; Finances in ChatGPT expanding to Free and Go users in the US on October 2; mobile camera multi-page PDF scanning on iOS and virtual try-on plus Favorites on mobile and web, both October 1. On October 15, OpenAI's ChatGPT release notes added GPT-Live-1.1 and a smaller "Quip" variant to ChatGPT Voice, with new voice options for French, German, Hindi, Japanese, and Korean, plus Mexican Spanish. None of these five entries mention dots by name — they're confirmed OpenAI announcements, just not ones that resolve anything about dots specifically.
Why the confirmed list matters even without mentioning dots
It would be easy to read the confirmed changelog entries as irrelevant to this piece precisely because none of them name dots. That's the wrong conclusion. A company shipping Locked Chats to every tier, widening financial-data access, and adding voice languages in the same three-week stretch is a company spending real engineering effort on trust, reach, and interaction surfaces — the kind of groundwork that tends to precede, rather than substitute for, a harder announcement about the product actually built on top of those foundations. Locked Chats specifically touches on the trust question dots raises directly: if a user can lock a chat against being read or modified by anything else running in their account, that's relevant to how comfortable they should feel granting a dot standing access to the rest of the account.
What's reported, but only by secondary sources
Two claims worth flagging precisely because they conflict. One guide dated around October 1 described dot creation as desktop-only, with mobile support described as coming later. A separate piece dated around October 13 reported that the ChatGPT mobile app can now create a dot, while mobile web still can't. Both are secondary sources, not OpenAI's own documentation, and they don't agree with each other on the state of mobile dot creation as of the dates each was published. Rather than pick a winner, the honest read is that mobile access to dot creation was in flux across the first two weeks of October and the current state isn't settled in the sources available.
Separately, reporting from Releasebot, dated roughly October 12-16, describes iOS widgets and Shortcuts integration for dots. That claim does not appear in OpenAI's own changelog in the pages checked for this piece — it should be treated as unconfirmed and single-sourced until OpenAI's own release notes corroborate it.
What OpenAI genuinely hasn't addressed
This is the real list, and it hasn't shrunk since launch:
What happens when the usage grace period ends. The first dot's usage didn't count against plan allowances for roughly the first month after launch. That window closes in late October, and OpenAI hasn't published what replaces it — not a discount schedule, not a new counting mechanism, nothing.
What additional dots cost. Beyond the first free dot, OpenAI has said more dots and narrower "specialist" dots will exist, without saying what either will cost. Three weeks of silence on this is now long enough to be a pattern rather than a launch-week gap.
Whether a dot can use a third-party plugin yet. A third-party guide flags this as unconfirmed, pending testing, rather than resolved either way — worth its own closer look.
Whether GPT-Live-1.1 touches dots at all. The October 15 voice release notes say nothing about dots. Any claim that this specifically upgrades how you talk to a dot is an inference from the fact that dots were announced with an audio interface "coming soon" — not a confirmed connection.
Three weeks is long enough to notice a pattern: OpenAI is shipping broadly, on a near-daily cadence, in every direction except the specific open questions dots users actually have.
The voice question specifically
GPT-Live-1.1's October 15 release notes are worth a closer, separate look because of how directly they intersect with something OpenAI said about dots at launch: that a phone and audio interface were "coming soon." Six new voice languages and a smaller "Quip" variant landing in ChatGPT Voice five weeks after that promise is suggestive timing. But suggestive timing is exactly the trap this series has tried to avoid walking into elsewhere — with Finances in ChatGPT's expansion and GPT-6.1 Sol's pricing, both cases where a plausible connection to dots exists without OpenAI confirming it. The same discipline applies here: GPT-Live-1.1 is a real, dated, confirmed OpenAI release. Its connection to dots' promised audio interface is an inference a reasonable person could make, not something OpenAI's own release notes state.
Why this matters more than it looks
None of these gaps are damning on their own — a month is a short runway for a company to finish pricing a genuinely new product category. But taken together, they describe a product that launched with its permission model and sandboxing reasonably well explained, and its economics and some of its basic cross-platform behavior still undocumented. The regulatory wait in the EEA, UK, and Switzerland is its own, separate open item — worth its own treatment rather than folding into this list, since it's a rollout-speed question rather than a product-definition one.
Part 42 of 47 in the ChatGPT Dots series. Previously: GPT-6.1 Sol vs GPT-6 Astra: Which Model Should Run Your Dot?. Next: The EEA, UK and Switzerland Wait Continues: Regulation vs Rollout Speed. Browse more builder insights or explore AI skills at aiskill.market.