In September 2026, OpenAI paused new sign-ups and upgrades to its $200-a-month ChatGPT Pro plan. Demand for GPT-6 Astra had outrun capacity. I recorded my video while the pause was still on. Since then OpenAI has reopened the plan, changed how much usage new subscribers get, and added a $500 tier.

This article covers what happened, where the Pro plans stand as of late October 2026, how to pick a plan without overpaying, and how to make your allowance last longer. Pricing and limits have changed twice in about a month, so treat the details here as a snapshot and check OpenAI’s pricing page before you buy.

Prefer to watch? See the full walkthrough on YouTube. I walk through the whole change on screen.

What happened to the $200 ChatGPT Pro plan

OpenAI released GPT-6 Astra on September 3, 2026. Within a week, heavy agentic and coding workloads were using far more compute than expected. On September 10, OpenAI stopped taking new sign-ups and upgrades to the $200 Pro plan (listed at the time as Pro 20X). Thibault Sottiaux from the Codex team announced the pause on X, and OpenAI’s help center confirmed it.

A few things stayed the same during the pause:

• People already on the $200 plan kept their subscription and their usage.
• The $100 Pro plan, Plus, Business, Enterprise and the API stayed open to new customers.
• OpenAI framed the pause as temporary while it added capacity.

In the video I explain my own situation. I moved from Plus to the $100 Pro plan because Plus credits ran out within minutes of Astra work. Then I burned through my Pro allowance in about five hours. That’s one person’s usage, not a benchmark, but it shows how quickly long agent runs eat into an allowance. I also show the pricing page as it looked during the pause on screen in the video.

You’ll see a lot of big numbers attached to this story on social media: daily spend figures, agent counts, view counts. I couldn’t verify them, so I’ve left them out.

Where the ChatGPT Pro plans stand now

The pause has ended. According to OpenAI’s help article on Pro tiers, here’s the current setup:

Plan	Price	What’s different
Pro 100	$100/month	Entry Pro tier with GPT-6 Pro in Chat, plus Work and Codex
Pro 200	$200/month	More usage than Pro 100. New subscribers get a lower allowance than the original Pro 20X plan
Pro 500	$500/month	Highest allowance and access to Ultrafast mode

Two details matter if you’re deciding right now:

• Grandfathering. Subscribers who were active on the $200 plan between September 22 and September 29 (10am PT) keep the earlier allowance through October 29, 2026. After that, expect the current allowance for your tier.
• Monthly billing only. Pro plans are billed monthly, so you can move up or down between billing cycles instead of locking in for a year.

Plus subscribers can use Astra in ChatGPT Work and Codex with limited usage, and can buy extra credits. GPT-6 Pro in regular Chat is reserved for Pro, Business and Enterprise. Free and Go plans don’t include Astra.

How to choose the right plan

Don’t start with the most expensive tier. Here’s the order I’d work through it:

1. Map where you use Astra. Chat questions, ChatGPT Work tasks and Codex coding runs draw on your allowance at different rates. Long autonomous runs cost the most.
2. Run a normal week on your current plan. Note when you hit limits and what you were doing at the time.
3. Try credits before a tier jump. If you only run out occasionally and your plan offers extra credits, buying a few can cost less than moving up a tier.
4. Move up one tier at a time. Go from Plus to Pro 100, then Pro 200. Pro 500 only makes sense if you need Ultrafast or run agents for most of the day.
5. Re-check after October 29. If you’re grandfathered, your usage on the $200 plan may drop after that date. Watch your first week and decide again.

If you’re moving onto Codex for real projects, AI Income Lab has Codex tutorials and templates, so you spend your allowance on the build instead of on trial and error.

How to make your allowance last longer

Most people run out of usage because of settings, not because their work needs it. These changes make the biggest difference.

Use medium reasoning effort by default

OpenAI’s model guidance recommends medium reasoning effort for most agentic coding and research work. Save high for hard debugging and long, multi-step tasks. Use the extra-high settings only when you’ve seen lower levels fail. In the video I mention that I keep Codex on medium for exactly this reason. Higher effort means more thinking tokens, and those count against your limit.

Be careful with Fast and Ultrafast

Speed modes cost more. OpenAI says Fast mode uses your allowance at 2.5 times the normal rate, and Ultrafast (Pro 500 only) at 8 times. Turn them on for a specific task that needs a quick answer, then turn them off again.

Match the model to the job

Not every task needs the flagship model. Renaming files, writing tests for simple functions and drafting short emails all run fine on lighter models. Keep Astra for planning, complex refactors and agent runs where quality really matters.

Keep tasks small and scoped

A single prompt like “build my whole app” can run for an hour and burn through a big piece of your allowance. Break the work into steps: plan, scaffold, one feature, test. You’ll catch problems early and spend less on retries.

Here’s a simple pattern that works well. Plan the task at medium effort, review the plan yourself, then let the agent build one piece at a time. Only switch to high effort when it’s stuck on the same bug twice.

If you’d like a ready-made setup for this, the AI Income Lab courses walk through coding workflows and prompts you can reuse.

Common mistakes to avoid

• Assuming the pause is still on. It isn’t. Pro 200 is open again, but new subscribers get a smaller allowance than before.
• Canceling a grandfathered plan to switch later. If you cancel and come back, you’ll get the current allowance, not the old one.
• Leaving reasoning effort on the highest setting. It’s the fastest way to hit a limit with little quality gain on routine work.
• Leaving Fast or Ultrafast on. At 2.5x and 8x usage, an all-day session adds up quickly.
• Expecting Plus to include GPT-6 Pro in Chat. Plus gets Astra in Work and Codex only, with limited usage.

FAQ

Is the $200 ChatGPT Pro plan available again?

Yes. OpenAI paused new sign-ups on September 10, 2026, then reopened the plan. New subscribers get a lower allowance than the original Pro 20X plan offered.

Why did OpenAI pause the plan?

OpenAI pointed to demand for GPT-6 Astra, especially heavy agentic and coding use, which needed more compute than it had available at the time.

What happens to grandfathered Pro 200 subscribers?

If you were active between September 22 and September 29 (10am PT), you keep the earlier allowance through October 29, 2026. After that, your tier’s current allowance applies.

Is Pro 100 enough for Codex work?

It depends on how long your runs are. For focused sessions at medium effort, many people get by. If you run agents for hours every day, you’ll likely need Pro 200 or extra credits.

Can I use GPT-6 Astra on the Plus plan?

Yes, but only in ChatGPT Work and Codex, and with limited usage. GPT-6 Pro in regular Chat requires Pro, Business or Enterprise.

Conclusion

The $200 plan pause was short, but it changed the deal. The plan is back, with less usage for new subscribers and a $500 tier above it. Before you upgrade, run a normal week on your current plan, drop reasoning effort to medium, and keep speed modes for the tasks that need them. Then move up one tier only if you still hit the limit.

Your next step: open your usage settings, note your reset window, and check your default reasoning effort in Codex today.

If you’d rather not figure this out alone, join AI Income Lab for the courses, templates and help that go with these builds.