On September 3, 2026, OpenAI released GPT-6 Astra, a new flagship AI model that replaces the GPT-5.6 family (Sol, Terra, and Luna) at the top of ChatGPT's lineup. If that sentence means nothing to you, don't worry — here's what actually changes for someone who just wants ChatGPT to keep working the way it does.
Who gets it, and when. OpenAI is rolling Astra out in stages rather than switching it on for everyone at once. It went first to a small group of cybersecurity partners, and is now rolling out over the following days to ChatGPT Plus, Pro, Business, and Enterprise subscribers, followed by developers using OpenAI's API and business customers on Amazon Web Services. If you use the free version of ChatGPT, you do not get Astra — for now, the free tier keeps running on the existing models. Unlike July's GPT-5.6 launch — when free and Go accounts got a new default model (Terra) on the same day paid users got the flagship Sol — this time there's no equivalent same-day upgrade for the free tier.
What's actually new. Astra is designed to be significantly better at multi-step tasks: reading a spreadsheet and reasoning about it, browsing a website to find specific information, writing and checking code, and carrying out instructions that involve several steps in a row rather than a single question and answer. OpenAI is also describing it as able to operate a computer somewhat independently — for example, working through a task across several apps or browser tabs with less hand-holding than before. For most everyday chatting, emailing, and quick questions, you likely won't notice much difference day to day.
Why security researchers are paying attention. OpenAI ranks its own models against an internal safety framework that tracks how much a model could help with dangerous tasks, including cyberattacks. Astra is the first OpenAI model to cross into what the company classifies as its "Critical" cybersecurity risk category — not because it's unsafe to use for normal purposes, but because it's now capable enough that OpenAI added extra safeguards and briefed cybersecurity partners before the wider release. For a regular ChatGPT user, this doesn't change anything about how you use the app; it's the reason OpenAI staged the rollout the way it did.
What it costs. If you already pay for ChatGPT Plus, Pro, Business, or Enterprise, Astra is included in your existing subscription — there's no extra fee to access it once it reaches your account. OpenAI is also offering optional credit top-ups for people who want to use it more heavily than their plan normally allows, similar to how add-on credits already work for image generation.
What changes for you, specifically. If you're on a paid plan, you don't need to do anything — Astra becomes available in the model picker inside the ChatGPT app over the next few days, the same way Sol, Terra, and Luna appeared automatically in July (see our guide on Sol, Terra, and Luna explained if those names are unfamiliar). If you're on the free plan, nothing changes for now, and there's no need to pay to "unlock" Astra through a link or app you find outside chatgpt.com — Sol, Terra, and Luna explained and is ChatGPT free? both cover what the free tier includes today. As always around a big AI launch, watch out for posts and ads claiming to offer early or "premium" access to a new model in exchange for payment or your login details — that's never how official OpenAI updates work; see our guide on spotting a fake AI app if something looks off. If you're already weighing whether a paid ChatGPT plan is worth it, our guide on ChatGPT Plus, is it worth it? covers what you get for the money.
Sources
- OpenAI unveils GPT-6 Astra with new cybersecurity guardrails — CNBC
- OpenAI releasing major upgrade to ChatGPT and Codex with GPT-6 Astra — 9to5Mac
- OpenAI's GPT-6 Astra launches today — 9to5Google
- OpenAI debuts GPT-6 Astra with new security measures — NBC News
- OpenAI rolls out GPT-6 Astra model with added cyber guardrails — Bloomberg