OpenAI’s new subscription plan is released! 5x and 20x packages have been removed

Abstract:

Dear subscribers, there seems to be something wrong with OpenAI’s package subscription pricing. On September 27, some users discovered that OpenAI had performed a "quiet operation" on ChatGPT's pricing page. The original Pro subscription seems to be different.

To put it simply, the $100/month and $200/month Pro plans are now called

"Pro Standard" and "Pro More" respectively.

5x and 20x packages we have been accustomed to, have all been deleted

The current usage description in the "Pro Standard" and "Pro More" packages only contains a vague sentence

"More usage than Plus"

Previously, the two levels of ChatGPT Pro had clear quantitative commitments: $100/month corresponds to 5 times the Plus usage, and $200/month corresponds to 20 times. This is the most direct decision-making basis for users to choose a subscription level. Now, those numbers are gone.

Interestingly, OpenAI still retains the old usage multiplier instructions in its own documentation. There was a "fight" between the pricing page and the documentation.

On September 25, OpenAI’s public openai/codex code repository merged a PR (#47971) titled: “Add Pro Max plan support and update Pro display names”. This 24-file change revealed three things:

First, a new promax solution type is added to the code, and the external display name is "Pro (Max)".

Second, the internal naming of existing plans has been adjusted simultaneously: the original prolite is changed to the display name "Pro" (i.e., Pro Standard on the pricing page), and the original pro is changed to "Pro (More)" .

Third, no specific price is written in the code. But earlier leaked in-app strings revealed that

Pro Max would be priced at $500/month

This is also the possible reason why the usage multiple is erased from the pricing page: the old 5x / 20x system is not enough to accommodate the third level, and OpenAI needs a new presentation framework to distinguish between the three Pro plans.

The leaked description of Pro Max is "Fastest Work and Codex", which points to two directions: the fastest inference speed and complete Codex programming capabilities.

There are unconfirmed reports that Pro Max may run on Cerebras' inference hardware to provide significantly higher response times than existing solutions. Some media outlets mentioned this statement but said it had not yet been confirmed by OpenAI.

The product logic behind this is clear: GPT-6 Astra's computing power demand is so huge that it forced the $200 plan to suspend registration, and OpenAI needs a new "ceiling" solution to achieve tiered supply of computing power. Users who are willing to pay $500 get the best resources, and the experience of users with $100 and $200 will not be degraded due to competition for computing power.

However, this also raises a question: When OpenAI simultaneously erases the usage multiplier commitment and launches a new high-price plan, will the rights of existing Pro users be "implicitly downgraded"? The vague statement "More usage than Plus" gives OpenAI a lot of room for flexible adjustment.

Reaction from the community was polarizing.

Some users think this is normal product stratification. The reasoning cost of AI models exists objectively. The more powerful the model, the greater the computing power it consumes. It is reasonable to provide high-priced solutions for heavy users. $500/month is not cheap for individual users, but it may still be a bargain for teams and enterprises that use ChatGPT to replace some of their manpower. Moreover, OpenAI’s adjustment is likely to be considering a price reduction.

Another area of ​​user dissatisfaction centers on transparency. Deleting specific usage commitments and replacing quantitative indicators with vague statements is usually not a good signal in the industry. Voices of "cutting leeks" and suggestions to unsubscribe have already appeared on Twitter.

DevDay is coming soon, will Pro Max be released directly?