Sunday morning, and instead of doing literally anything else I'm reading a changelog entry about weekly token allowances. This is what my life is now.
The entry says Anthropic is permanently raising standard weekly limits in Claude Code by 25% for Pro, Max, Team and seat-based Enterprise, starting September 14. Good news. Except the temporary 50% boost that's been running since May 13 ends on the same day, so for anyone using Claude Code today the actual move is downward. Anthropic's own wording: "Compared to today, this works out to a 17% reduction in weekly limits on Claude Code."
The maths isn't complicated. Baseline 100, boosted 150, new permanent floor 125. Up against the old baseline, down against your Tuesday.
The part I want to be fair about
They said the 17% out loud, in their own announcement, without being asked.
Sit on that for a second, because it's rare. The normal corporate move is a post titled "Increasing Claude Code limits by 25%" with the boost expiry buried in a footnote, or more likely not mentioned at all. Then you find out three weeks later when your Thursday session dies early and you spend an hour convinced you broke something. They published the subtraction instead. That deserves more credit than it'll get, because far fewer people notice the honesty than notice the cut.
Same week they cancelled the scheduled Sonnet 5 price increase and made $2/$10 per million tokens permanent instead of going to $3/$15 on September 1. That's a straight money-in-your-pocket win and it got maybe a tenth of the attention the limit cut did. Which tells you something about all of us, me included.
So what's actually wrong with it
The mechanism is the problem here, and it's a self-inflicted one.
The boost landed on May 13. Then it got extended. Then extended again. Four separate times over the summer, each time shoving the cliff a few weeks further out. By the third extension nobody was treating it as temporary any more. Everyone built their week around 150 and quietly started assuming 150 was just the number.
That's what happens with a promotion you renew four times. It stops being a promotion and turns into the baseline in everyone's head, and then going back to an honest permanent number feels like something got taken away. You can be completely straight with people and still end up in a hole you dug yourself.
I don't think anyone was being sneaky. It reads like a capacity call that got extended whenever the numbers allowed, which is a sensible way to run infrastructure and a terrible way to set expectations.
What actually bugs me
Run the numbers on a heavy week through Max at list API rates and the subscription looks almost silly. The $200 plan isn't where anyone is getting squeezed. If anything the heavy users are the ones being subsidised, and I'd rather that stayed solvent than watch it get yanked entirely in a year.
So no, this isn't a post about wanting cheaper tokens.
It's about not knowing what your number is. Build a real working process on top of an agent, a team habit or an actual automated pipeline, and you need to know your weekly capacity well enough to plan against it. A ceiling that moves because a promotion lapsed is useless for staffing, useless in a proposal, and barely explainable to a colleague without drawing a diagram.
That's the real cost, and it's bigger than 17%.
This is the second squeeze this quarter
Worth remembering the programmatic side already moved. Agent SDK usage, claude -p, GitHub Actions, anything non-interactive, came off the subscription and onto its own metered credit pool at standard API rates in June. Pro gets $20 of credits, Max 5x gets $100, Max 20x gets $200, and once you're through them you're paying list price on top of a subscription you already paid for.
I got the logic then and I get it now. People were hammering enormous automated workloads through a flat fee and that was never going to hold. (Anyone doing it knew exactly what they were doing.)
But stack the two changes and the shape is obvious. The era where a subscription quietly covered whatever you threw at it is over, and it ended inside a quarter. If your plans assumed otherwise they need redoing, and waiting doesn't make that surprise any cheaper.
What I'm doing before the 14th
Measuring. That's it, really.
Most of this year the ceiling was far enough away that there was no reason to care what my actual burn looked like, so I never wrote it down. That's the bit that bites you. You don't find out you were living at 150 until the day it's 125.
So for the next two weeks I'm keeping a real log, so that on September 14 I know whether I'm in the group where 17% is a rounding error or the group where it's a Wednesday that ends at 3pm. Then the batch and scripted work moves off the interactive plan where it belonged all along.
And if you're running a team on this, do that now rather than on the 15th when someone asks why the sprint stalled.
If Anthropic ships the "more visibility and control of your usage" they hinted at in the same announcement, half of this stops mattering, because what I actually want is to know where I am against the number I've got rather than a bigger one.
