Saturday, September 12, 2026

News

OpenAI Pauses Sales of Priciest ChatGPT Plan After GPT-6 Astra Surge

MarketPatryk Raba

OpenAI has suspended new sign-ups for its $200-a-month Pro 20X subscription, saying users on that tier are putting the heaviest strain on its infrastructure since the launch of the GPT-6 Astra model.

Contents
  1. What exactly was closed
  2. Why this plan specifically
  3. The scale of the problem and what's next
  4. A pattern of earlier restrictions

OpenAI has stopped selling new subscriptions to its most expensive ChatGPT tier. The reason is demand for the GPT-6 Astra model, which has put more strain on the company's servers than any previous release.

The news was first reported by Telepolis.pl, citing Sottiaux's posts on X. The company isn't restricting access for all ChatGPT users, only closing the door to new subscribers of its priciest tier, aimed at heavy users of reasoning models and coding tools.

What exactly was closed

The Pro 20X plan costs $200 a month and grants expanded usage limits on OpenAI's most powerful models, including GPT-6 Astra. As of now, it can't be purchased by new subscribers or upgraded to from the cheaper $100 Pro plan, Go, or Plus.

People who already had a Pro 20X subscription keep their access unchanged. The restriction also doesn't affect access to OpenAI's models via the API, which companies use to build their own applications.

Pause subscriptions to our $200 Pro - Thibault Sottiaux, OpenAI

Why this plan specifically

According to Sottiaux, Pro 20X users are putting the greatest strain on OpenAI's systems because they make heavy use of GPT-6 Astra and its reasoning modes, which consume the most compute per query. Limiting the inflow of new customers to this segment is meant to let the company maintain service quality for people who already use it.

put the most strain on our systems - Thibault Sottiaux, OpenAI

The scale of the problem and what's next

Sottiaux gave no timeline for when Pro 20X sales will resume. He said only that OpenAI is increasing available compute capacity as fast as it can. That phrasing is not the same as a firm deadline or a guarantee of a quick fix.

Telepolis.pl notes unofficial reports from users complaining about a drop in the quality of GPT-6 Astra's answers in recent days. One possible explanation, flagged in the article as speculation rather than a confirmed fact, is model quantization or a shortened "thinking" time before responding, which would ease the load on infrastructure at the cost of quality.

A pattern of earlier restrictions

This isn't the first time in recent weeks that OpenAI has responded to infrastructure strain by changing access to paid ChatGPT and Codex plans. The company previously reset usage limits for paying Work and Codex plan users, and reinstated a five-hour limit window for the Plus plan.

For Polish companies and freelancers using OpenAI's priciest plan, this means adding another team seat with Pro 20X access isn't currently possible by buying a new subscription. Existing users won't see any change in access for now, though unofficial signs of a possible drop in Astra's answer quality should be treated cautiously until OpenAI addresses them.

The episode shows that even a company with the largest compute infrastructure among AI labs can be caught off guard by how fast users flock to a new, more expensive-to-run model. GPT-6 Astra arrived in ChatGPT as a new-generation reasoning model, and its most compute-intensive modes turned out to be more popular than the company anticipated when planning server capacity.

Share: