天塌了！兄弟们！
老金我一直推荐网页版的6 Pro，在我心里，它就是最强推理模型，没有之一。
慢是真的慢，但我愿意等，因为它会补背景、追问，把一个问题翻来覆去想周全。
比起赶紧拿到一个答案，我更喜欢跟它慢慢聊明白。
我没事儿也会找ChatGPT聊天，让它多了解我。
等需要它帮我填信息的时候，Memory这个记忆功能，就能把以前记住的内容调出来，不用每次从头介绍自己。
这些聊天在后面真要做事的时候，它知道的背景越多，我要补的话就越少。
现在，OpenAI要把这两件事放在一起算账了。
官方帮助页已经写明，12月起，Plus、Pro和Business用户的ChatGPT与Codex共用额度，按Token计量，Instant快速聊天不扣这份额度。
Token可以理解成模型处理和生成信息的计量单位，不是简单数你发了几条消息。
用更高级的模型、做更复杂的任务，通常消耗也更多。
目前Work和Codex已经共享用量，12月还要把Chat里会扣额度的聊天合进来。Chat和Work的界面则是未来几周先小范围测试合并，Codex仍保留独立体验。
偏偏在这之前，我才刚松了一口气。
9月30日收到的旧邮件说，我这个每月200美元的Pro，Work和Codex额度原本要从Plus的20倍降到10倍，10月30日开始执行，价格不变。
钱照交，量打对折，这可不是什么半价优惠，当时我还吐槽过。
但北京时间11月1日凌晨，新邮件又说，从10月29日起保留这部分额度，之前宣布的这项削减不再执行。
原本要少掉的一半保住了，我当然高兴。可转头看见12月的新规则，又得重新琢磨，聊天也从同一份额度里扣，以后这份量究竟够不够？
合并后的具体额度、12月哪天切换，官方都还没公布。
但这次变化，确实把一个我原来不怎么想的问题摆到了面前。
以前是不懂就问，后面可能变成不懂先忍忍。
可站到OpenAI这边看，这套规则又不是完全没道理。
同样在聊天框里，一句简单问答，和带着一堆材料让模型认真推敲，背后做的工作并不一样。
再往前一步，让它直接把事情做完，聊天和干活的边界就更难按入口划了。
从计量的角度，按实际处理的信息和任务消耗来算，比只数发了几条消息，更容易把不同用法放到一起。
Claude的聊天、桌面端和Claude Code，本来就已经共用额度。
所以，换Claude并不能躲开这件事。何况我还被它封过3个充值号，到现在没啥好方案，不敢用。
我能理解厂商需要把消耗算清楚，但用户付月费，也有用户的算盘。
现在若要把聊天和后面的任务一起盘算，就得想，这个问题值得用最好的模型吗？
再问一轮，还是把额度留给后面？
本来花钱是想省心，别到最后，省下来的全是自己想问的话。
当然，Instant还在，而且不扣共享额度。
可我喜欢6 Pro，恰恰是因为它愿意慢下来，把事情想周全。
能接着说话，和能得到我想要的回答，是两回事。
更何况，一个问题有多难，有时也是问下去才知道的。让用户事先分清哪句话值得认真想、哪句话随便答就行，并没有听起来那么简单。
写在最后
官方预计，多数用户不必比现在更频繁地等额度重置，真能做到，老金当然高兴。
6 Pro的价值，对我不只是把已经想清楚的任务做完。
还有我没想清楚的时候，可以把问题拿出来，跟它一起想。
我愿意多跟它聊，让它记住我，也是为了以后少一点解释，多一点默契。
希望12月以后，我打开那个越来越了解我的聊天框，还是可以从一句没想好的话开始。
谢谢你读我的文章。
如果觉得不错，随手点个赞、在看、转发三连吧🙂
如果想第一时间收到推送，也可以给我个星标⭐～谢谢你看我的文章。
开源知识库（实时更新 交流群）：
my.feishu.cn/wiki/OhQ8wq…
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
The sky is falling! Brothers!
Kim, I've been recommending six Pros of the web page, and in my mind, it's the strongest reasoning model, none of it.
Slow is really slow, but I'm willing to wait because it fills in the background, asks questions, and turns a question around.
I'd rather talk to it than get a quick answer.
I'll talk to ChatGPT to make it know me better.
When it's needed to fill out the information, it's the memory function of memory that pulls out what you remember, without introducing yourself from scratch.
The more it knows the background, the less I have to add.
Now OpenAi's going to put these two things together.
The official help page already states that since December ChatGPT, a user of Plus, Pro and Business, and Codex, shares quota, which, as measured by Token, Instant's fast chat does not exclude quota.
Token can be understood as the unit of measure for model processing and generating information, not simply a few messages.
More sophisticated models and more complex tasks are usually consumed.
Work and Codex are now sharing usage, and in December the Chat will be joined by a chat with quota. Chat and Work's interface is a combination of small-scale tests in the coming weeks, and Codex retains an independent experience.
Before that, I was relieved.
Old mail received on 30 September stated that my $200-per-month Pro, Work and Codexquota would have been reduced from 20 to 10 times the Plus and would have been implemented on 30 October at constant prices.
It's not half-priced. I've done it.
However, in the early hours of 1 November Beijing time, the new mail added that this part of quota would be retained as from 29 October and that the previously announced reduction would not be implemented.
I'm glad I saved half of it. If you turn around and see the new rules of December, you'll have to rethink them, and you'll have to take the chat from the same quota.
The actual quota, the date of December, has not been officially announced.
But this change does bring to the fore an issue that I did not think of.
It's been a while since I didn't know what to say.
But it doesn't make any sense to come to Open AI.
Also in the chat box, a simple question-and-answer line is not the same thing that is done behind the model with a bunch of material.
One more step forward, so that it can finish things directly, and the boundaries of chatting and working are harder to cross.
From a measurement point of view, the actual processing of information and task consumption makes it easier to combine different usages than to send a few messages.
Claude's chat, desktop and Claude Code are already sharing quota.
So Claude can't get away with this. Besides, I've had three top ups sealed by it, and I don't have a good plan.
I understand that manufacturers need to calculate consumption clearly, but users pay monthly fees and there are user calculations.
Now, if we're going to think about talking together with the tasks that follow, we have to think, is this a question that deserves the best model?
Another round, or leave quota behind?
The money was meant to save your heart, not to end up saving everything you wanted to ask.
Of course, Instant is still there, and the sharing of quota is not excluded.
But I like Pro 6, precisely because it's willing to slow down and think things through.
It's different to be able to talk and get the answers I want.
Moreover, how difficult a problem is, sometimes it is to ask again. It is not as simple as it sounds to allow users to identify in advance which words deserve serious thought and which words to answer.
At the end.
Officially, most users do not have to wait more frequently for quota to be replaced than they are now. This is true.
Pro's value is not just for me to finish what I've already thought out.
And if I don't figure it out, I can put it out and think with it.
I'd like to talk to it more, so that it can remember me, and so that in the future, there's less explanation, more understanding.
Hopefully, after December, I'll open that chat box that's getting to know me better, or I'll start with something I didn't think of.
Thank you for reading my article.
If you think it's good, you can order it, read it, forward it.
If you want to get a delivery first, you can give me a sign. Thank you for reading my article.
Open-source knowledge base (real-time update exchange group):
my.feishu.cn/wiki/OhQ8wq…
Auto-fetched and translated by AI Atlas. All rights belong to the original author.