Last Call: Codex 200's $200 Credit Quota Is Halved as ChatGPT Gets a Major Update

  It's the first time in the history of the world.

It's only the last day until October 30!

OpenAI's $200 Pro subscription users can't sit down, and the collective countdown has begun these two days.

Starting tomorrow, the Codex and ChatGPT Work allowance will be reduced from 20 to 10 times as much as the Plus, and the GPT-6 Pro's weekly messages will be cut in half.

For a group of developers who have been working day and night in the codex, allowing AI to change the code and run tests at the same time, the tasks that could have been completed are now focused on quantitative searches.

It's a scandal that OpenAI has taken out $2,500 in compensation.

But the truth is, the whole network is open, and the wave that can really save the field is probably the unproduced GPT-6.1 Astra.

Just this morning, ChatGPT brought together "chat" and "work".

In the selection of models, it is also different from the previous one. If you want to use the Pro model, you have to first select the GPT-6 sol, and then you have to raise the reasoning intensity.

One wave is not flat, one wave is up again. Everyone is still the most concerned about what constraints this will have on the volume of traffic.

Tibo, the God of Replacement, immediately responded that the integration of ChatGPT has not yet been officially launched, and more information will be shared soon.

The dispute has been ongoing since the end of September.

On the eve of DevDay, OpenAI officially announced an adjustment to the ChatGPT Pro subscription system and reopened the previously suspended 200K subscription package.

There is also a higher price of $500 Pro packages, all of which are included.

For $200, older users receive a "transition period" of about a month.

That is, between September 22 and September 29, at 10 a.m., users who had a valid Pro 200 subscription can continue to enjoy the original quota.

And the monthly fee, or the $200, is no less.

Looking at some of the emails received by the affected users, you can see that the shrinkage is mainly two points.

The first knife, cut directly into the Codex and ChatGPT Work.

In the past, the $200 Pro package included 20 times the usage limit of ChatGPT Plus. After adjustment, only 10 times remained.

The second knife, cut to the GPT-6 Pro.

In the ChatGPT chat interface, the GPT-6 Pro used 200 messages per week.

In order to appease the old users, the OpenAI wave directly paid a compensation: 62,500 Codex points per person, an official discount of almost $2,500.

It sounds like a lot, but it has a "warranty period" after December 31, 2026, and it has not been fully cleaned.

After the adjustment, the Pro members directly created the "Three Brothers": 5 times the 100 knives, 10 times the 200 knives, and 500 knives if they wanted 25 times.

In addition, it also sends an ultrafast mode called Ultrafast, which can reach up to 8 times the speed.

I'm afraid of the unknown, I'm afraid of the unknown.

The Claude Max 20x is also a monthly 200 blade, which is directly equivalent to 20 times the power of the Pro, and is also calculated according to a five-hour dynamic window.

Most notably, the Anthropic family's top-selling Fable 5 and 5.1 are all directly synchronized within the Max.

Google's AI Ultra is priced at $199.99, which is one cent cheaper than the 200 knives, and marks a 20-fold upper limit on the AI Pro package.

Back to OpenAI, starting tomorrow, the same price is 10 times the price of Plus.

What's more, just one day before the deadline, OpenAI came up with another trick.

Just this morning, ChatGPT's input was updated, and a big wave of developers discovered that the top-of-the-line "chat work" switch had disappeared.

ChatGPT unified mode, the phone is online.

In July this year, ChatGPT Work went live, and the desktop was switched to a "chat/work" switch:

On the one hand, it's a question-and-answer chat, and on the other hand, it's a Work that can run for hours on end, delivering documents and forms directly.

Shortly after, Brockman, president of OpenAI, admitted in an interview that the new version of the app was a bit messy and promised not to have a separate Work label by the end of the year.

Earlier this month, Tibo also publicly announced that the two models would merge into one to reduce the complexity of use.

From a product standpoint, it's a good thing: you don't have to put any more buttons on that button and ask questions to get ChatGPT to work.

In this case, the developer's main concern is only one question: What is the quota after the merger?

It's worth noting that this Pro 200 is exactly the "Codex+ChatGPT Work" piece.

When the line between talking and working disappears, will the usual hand-to-hand conversation begin to draw from that half-shortened quota?

When the news of the merger came out, the discussion on Reddit surged to hundreds of likes, and some even started asking not to unsubscribe.

It's hard not to think too much about these two things.

Now the entire network is waiting for the GPT-6.1 Astra rescue, expecting it to roll over.

However, the reality is that the chances of it arriving before the Great Depression are slim.

On the day of Tibo's announcement of the quota, OpenAI immediately stepped on the brakes.

Saachi Jain, Head of OpenAI Security Systems, revealed that 6.1 Astra has not yet reached its release and alignment.

One is to live without oneself, the old love of Vietnam, and the other is to be untruthful and not be able to really interact with the users.

In internal testing, the GPT-6.1 Astra moves on its own without human consent, invokes potentially unsafe external tools, and is not always honest when reporting work.

Today, the competition for AI programming tools is entering a new phase.

For a heavy-duty developer, a package is worth nothing, not just to see how smart the model is.

The real question is, how much do I spend on $200 this month to make it work for me? This figure is equal to multiplying the capacity of the model by the quantity and dividing it by the amount of computing that needs to be burned for each task.

As we look deeper, the subscription model itself is changing.

The last 200 knives, more like a vendor's over-priced cafeteria, eat more and eat less at the same price.

However, for developers, the cost of switching tools is only one line of command.

If you're the guy who makes $200 a month, the most important thing is to get up now!