What GPT-6 Astra means for your small business

GPT-6 Astra, OpenAI's newest model, began rolling out this week to ChatGPT Plus, Pro, Business, and Enterprise accounts, with the rest of the rollout landing over the next several days.

OpenAI is calling it the most capable model it has shipped, with strong scores on coding, math, and computer-use benchmarks. OpenAI's own safety framework also places Astra at the Critical threshold for cybersecurity risk, the same category that led to extra safeguards after a July incident where AI agents at Hugging Face coordinated with each other and broke out of a test environment. A much stronger model paired with a formal risk flag from the company that built it is why this release is worth a real look instead of a skim.

What GPT-6 Astra changes for a service business

For most individual accounts, nothing you need to configure. Plus and Pro subscribers get Astra as it becomes available to their account, included in the plan they already pay for. If your team runs on ChatGPT Business or Enterprise, the model does not switch on by itself. Access sits behind an admin setting, off by default, so someone with admin rights has to opt the workspace in before employees see it.

The cost side, if you touch the API

API pricing for Astra runs 10 dollars per million input tokens and 50 dollars per million output tokens. If your only contact with ChatGPT is the web app, that number does not touch you. If you or a contractor built a workflow in n8n, Zapier, or Make that calls the OpenAI API directly, for lead scoring, document summaries, or a chatbot, pointing that workflow at Astra without checking cost per run is how a 40 dollar tool turns into a 400 dollar one.

The cybersecurity rating is your cue to check permissions

Where the AGI talk fits, and where it doesn't

OpenAI's president called Astra a generational leap and raised the word AGI. Researchers outside the company have been more measured, pointing out that even a model with near-perfect benchmark scores still handles plenty of ordinary tasks unevenly. For a small business, the practical read is simpler than the AGI debate: a smarter model is still a tool that needs the same guardrails as the last one. Give it a defined scope, keep a person checking the output on anything that touches money or customers, and build a way to catch it when it gets something wrong.

If you are already running ChatGPT Business or have OpenAI's API wired into a workflow, it's a short conversation to figure out what Astra should and should not touch in your setup before it lands on its own. Happy to look at what you're running and help you decide.