A 25% quota increase can still reduce next week’s capacity
Claude Code’s announced change compares with the standard allowance. Capacity planning also needs the comparison with the temporary allowance users currently receive.
Anthropic’s announced Claude Code change replaces a temporary 50% increase in weekly limits with a permanent 25% increase over the standard allowance from September 14. BleepingComputer’s August 29 report reproduces the clarification that this is roughly a 17% reduction relative to the temporary allowance. The two percentages use different baselines.
Set the original allowance to 100. The temporary allowance is 150. The new allowance is 125. Compared with the original, 125 is a 25% increase. Compared with 150, it is a reduction of 25 divided by 150, or 16.7%.
Both calculations are true. Only the second describes the transition a user experiences at the end of the promotion.
My rule for a capacity announcement is to show the old standard, the currently effective allowance and the next effective allowance together. One percentage cannot describe all three.
The planning baseline is the week you measured
Suppose I am planning a hypothetical coding workflow from a week observed during the temporary increase. The service handled a certain amount of work before hitting its limit. If I label that observation normal capacity, the forecast carries a hidden condition into the following month.
I would attach the applicable allowance and date range to the measurement. Then I would identify which future weeks have the same conditions and which need a revised plan.
That does not mean multiplying task output by 0.833 and calling the result precise. Usage limits do not necessarily translate into a fixed number of completed tasks. Prompt length, model selection, effort and tool activity can change consumption. The ratio describes the allowance transition, while actual task capacity still needs observation.
The practical concern is headroom. A team operating close to the temporary ceiling has a different exposure from one using half its allowance. The same product announcement can require immediate workload changes for one and no action for the other.
Write the transition as a dated row
For the hypothetical team, I would record the current allowance, its expiry, the announced replacement and the workflow’s recent usage. Then I would identify the tasks that can wait, the work that needs a separate capacity source and the point at which the team should revisit its commitment.
I would show both baseline comparisons in the internal note. That prevents a discussion about whether the change is really an increase or a reduction from replacing the operational decision. It is an increase relative to one state and a reduction relative to another.
The announcement’s original deleted wording is unnecessary to make that decision. The surviving clarification and the arithmetic provide enough information. Speculating about motive would add heat without improving the capacity plan.
This also gives finance a clearer record when the subscription price stays constant. The relevant event is a dated change in the amount of service available under the plan, with a separate assessment of how that affects the work.
Promotions can be useful. Their expiry should simply remain visible in the evidence used to promise future delivery.
A capacity forecast needs the allowance that will apply when the work runs, plus the baseline under which the forecast was measured.
