Finally!Claude Code's quota is up and he's no longer cutting your code.

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

In the end, Claude Code did his job.

Just yesterday, an Anthropic developer account @ClaudeDevs announced: "Claude Code hit the five-hour limit and stopped cutting a knife.

Now, it will find a decent stop, take this step and then stop.

It's a change that's blown up in the comments section.

Anyone who has ever used Claude Code knows how disgusting this knife is.

It's not the disgust of being interrupted, of repeating a word.

It's like you've changed the signature of a function, the place of the call hasn't changed, the test is half run, and Claude suddenly stops.

The developer @tochukwudev says most directly: This is his biggest problem with Claude Code, which is that every time it's cut, it's left half-cut and needs to be handled a little bit.

In order to escape this sword, people practiced all sorts of earthly methods.

@PaineofGlass said he used to rush to the limit to have Claude split up a few disks, or simply raise the limit on excess spending.

You spend money on your goods, not to spend too much, but to avoid dying too hard to see.

This new feature is called Wrap-Up Allowance.

It's called the "finishing line".

The mechanism is simple: the moment you hit the five-hour limit, the system pulls out a small fixed limit from your weekly allowance, allowing Claude to stop doing this step.

The user interface will display a line that says "Usage limit reached · wrapping up".

The teacher asks you to write the sentence that you are writing and turn it over.

Sounds reasonable, doesn't it?

But before that, Anthropic's approach was: When the bell rings, the pen is pulled straight out of your hand, and you write wherever you want, whether you're writing half of the scroll or just starting the pen.

Anthropic finally knows that it was a bug-level experience before.

But don't get too excited about it, the official help center is clearly drawing the border:

First, if the wheel is running.You send a new message after hitting the line, you're blocked, and you can only wait for the amount to be reset. Do you want to fill out an instruction while closing the window?

Secondly, no extra weight.Anthropic doesn't give you more than a minute, it just changes how to stop after that moment.

Thirdly, there is no guarantee of completion.Finish the quantity with light, this step is done and the next step is no longer started. You can also interrupt it manually at any time.

This quota cannot be applied for, can't be mined, and can't be transferred next time, and is automatically effective if the conditions are met.

It's not enough to start extra usage and run. The system uses the end quota first and then deducts your usage quota at the standard price, which is still restricted by consumption.

There's a hard-hitting classification here.

Pro users:One week at the most. Even if you don't use it all, you've already used it. Translate: Seven days a week, you only get one chance to "stop decently", and the remaining six times you hit the wall or get stabbed.

Max and Team Premium:Each time you hit the five-hour limit, you can use it, as long as the weekly limit is not exhausted.

Several other thresholds must be met: Claude Code must be upgraded to 2.1.277 or an updated version, the terminal, the IDE plug-in, and the desktop's Code page are all valid, but must be subscribed with Claude's account.

The account and the organization in which it is located have to exist for a period of time (this time is not officially public, and may be modified later).

You can't call with API Key or use Amazon Bedrock or Google Cloud Vertex AI.

How much is the final quota? The official only said that "it varies according to the package, there is a cover" and did not give a number.

Typical anthropic style: tell you something like this, but don't ask how much.

On September 14, Claude Code's temporary 50% weekly promotional allowance expired, replacing it with a permanent 25% increase.

According to the original quota, 100 is about 17% less than the maximum from 150 to 125 tonnes.

At first, Anthropic only said "permanently raise 25%" on X, was booed by users and posted to community notes, and then deleted and re-posted, only to honestly admit, "This is down 17% compared to now".

The weekly quota has been tightened and the final quota has been deducted from the weekly quota.

Anthropic didn't give a new quantity for a minute, and the change was purely how to stop using the quantity after that moment.

The comments on the X were warm and clear against the camp.

@DataDiscovered said that the editing process was the worst of Claude Code's failures in the middle, and that the end result was one of Anthropic's best repairs to date.

The most direct is @xDqtreats, which he says is changing from Codex because he likes it.

@nrajesh4u says he has been responding to this need for a long time, and it is estimated that thousands of people have mentioned the same thing.

It's okay to drink cold water.

@shamilkch believes that it makes sense for the next session to start from a good written checkpoint and end. Otherwise, this fixed end limit will just keep Claude quiet, which in essence is an overtime, just a good look.

Others were staring at the five-hour limit itself. @ysdevastro directly shouted: Can you give Max the five-hour limit?

Some people run out of the codex because of this feature, while others stay in the codex because of a five-hour restriction.

The two are the same people who write the same code, and the quota rules have long been a serious comparison when choosing developers.

In retrospect, this change is really small.

The five-hour window is not loosened, the weekly limit is not increased, and even the end-use limit is deducted from your original weekly limit.

But the comment section is a good one. Why?

Because the way AI works has changed.

In the past, people used a chatbot to ask a question and answer it, and the cost of repeating a question was close to zero.

But it's not the same with an agent like Claude Code, who reads several files at a time, changes several codes, runs a round of tests and changes, and runs a task for decades is normal.

Its state is not in a single answer, but in the middle of the whole series of interactions. Deciding which step stops you when you come back to pick it up is to continue doing it or to spend half an hour packing up leftovers.

So Anthropic doesn't have much to offer, just change the location of the stop, and everyone's welcome.

Because whenAIThe longer and more complicated it is, the more it stops where it stops, the more you can do it after it stops, it's no longer a question of "experience optimization".

It's the product itself.