Why can't I access GPT-6 Astra on my paid ChatGPT plan?

OpenAI released Astra in stages rather than all at once. Organisations in its Daybreak cybersecurity programme received it first, followed by enterprise customers, leaving Pro, Plus and Business subscribers waiting. OpenAI has said the broad rollout to ChatGPT subscribers and API customers will begin with Pro accounts, but has not given a firm date.

What is the banked reset OpenAI is giving ChatGPT subscribers?

OpenAI is granting one banked usage reset for every day a paid ChatGPT subscriber goes without GPT-6 Astra access, counted from 3 September 2026. A reset refreshes your usage allowance early rather than increasing the cap. They accrue automatically, so there is nothing to claim, though OpenAI has not yet specified exactly which allowance each reset refills.

Do free ChatGPT accounts get GPT-6 Astra?

No. GPT-6 Astra was announced for paid plans only, and free accounts stay on the existing default model. Free users are not part of this rollout and are not earning the banked resets, since those apply to paid plans.

Will updating the ChatGPT app give me GPT-6 Astra sooner?

No. Access is granted server-side against your account, so reinstalling or updating the app makes no difference. When your tier is switched on, Astra appears in the model picker in a new chat. On Business and Enterprise workspaces an administrator must enable it first.

How much does GPT-6 Astra cost on the API?

OpenAI lists Astra at $10 per million input tokens and $50 per million output tokens, with the fast mode charged at roughly double those rates. API access, along with Microsoft Azure and AWS Bedrock, is part of the same widening rollout rather than being available immediately.

OpenAI Apologizes for GPT-6 Astra Delay, Adds Daily Resets

Why Pro subscribers watched enterprises go first

Astra did not ship to the usual queue. Access opened first to organisations vetted through Daybreak, OpenAI's defensive-cybersecurity programme, which received the model with expanded permissions for security work. Enterprise customers followed. Pro subscribers — the $200-a-month tier that normally gets frontier models on day one — were left waiting behind both.

That inversion is what set the complaints off. Replies under OpenAI's own launch posts filled with subscribers pointing out they were paying the most and getting it last, and threads asking a blunt version of the same question ("how many people actually got access?") ran through the ChatGPT communities all Friday morning.

In a public post, Altman said sorry for the rollout, added that the company tries to make things right when it gets something wrong, and said broad access for API customers and ChatGPT subscribers should begin soon — starting with Pro. He would not commit to a date, saying only that he hoped people could use it over the weekend but could not promise it.

What the banked reset actually gives you

The compensation came from OpenAI's Codex engineering lead, Thibault Sottiaux, who said paying users get one banked usage reset for each day they spend without Astra access, counted from 3 September. A reset restores your allowance early rather than raising the cap, so a subscriber who has been locked out for three days by the time access lands would have three of them sitting in reserve.

One detail is not yet pinned down: OpenAI has not spelled out precisely which allowance the reset refills — the weekly Astra message budget, the wider plan quota, or both. Until the company says, treat the exact value as unconfirmed. What is confirmed is that they accrue automatically per day, so there is nothing to claim and no form to fill in.

How to tell whether you have it

The rollout is server-side, so reinstalling the app or updating it changes nothing. A few checks worth making before assuming you have been skipped:

Open the model picker in a new chat rather than an existing one. Astra appears there when your account is switched on.
Business and Enterprise workspaces ship the model turned off. An administrator has to enable it before anyone in the workspace sees it, which is a different problem from a slow rollout.
Free accounts are not in this wave at all and stay on the existing default model, so there is nothing to wait for on a free plan.
Sign out and back in once access is announced for your tier. Entitlement changes sometimes need a fresh session to appear.

Developers are in the same queue. Astra is due on the API alongside Microsoft Azure and AWS Bedrock as the rollout widens, at $10 per million input tokens and $50 per million output tokens, with the fast mode charged at double that. We covered the plan-by-plan split and the full pricing when the model launched, in GPT-6 Astra lands in ChatGPT for Plus and Pro users.

The second rollout OpenAI has had to apologise for

Altman also referenced the GPT-5 launch, when he acknowledged the company had badly mishandled parts of the release. That one ended with OpenAI restoring an older model people had lost and lifting limits on the newer one. This time the problem is narrower — the model works, it is simply not reaching the people who pay for it — but the pattern of shipping first and sorting out access afterwards is now familiar enough that subscribers recognised it immediately.

Nothing about the staged approach is unreasonable on its own. Handing a frontier model to vetted security organisations before the general population is a defensible call. Announcing it as a launch while the paying tiers wait is what turned a phased rollout into a complaint thread.

If you are on Plus or Pro, the practical position is this: you are accruing resets while you wait, Pro goes first when the gates open, and there is no action to take in the meantime. For everything else changing across OpenAI's consumer plans, see our coverage of unlimited text chats on the free tier and the teen mode parental controls.