If Claude Code’s weekly allowance felt tighter after September 14, there is a documented reason. Anthropic ended its temporary 50% weekly boost and replaced it with a permanent increase of 25% over the original baseline for eligible plans.
That leaves you with more weekly capacity than before the promotion, but less than you had during it. The difference is 16.7% below the promotional allowance. Five-hour limits were outside this promotion.
What changed on September 14
According to Anthropic’s official promotion explanation, the offer ran from May 13 through September 13, 2026, ending at 11:59 p.m. Pacific Time. The permanent 25% increase started September 14.
Eligible plans include Pro, Max, Team, and seat-based Enterprise. The promotion’s Enterprise eligibility specifically covered legacy seat-based users; Free plans and consumption-based Enterprise seats were excluded.
The temporary boost applied to Claude Code across its CLI, IDE extensions, desktop, and web interfaces. It did not raise the limits for other Claude products, including Claude chat and Cowork.
The arithmetic: 100 → 150 → 125
Use 100 as an illustrative starting allowance. These are index values, not an actual allocation of messages, tokens, or coding hours.
| Period | Illustrative weekly allowance | Compared with the original baseline |
|---|---|---|
| Before the promotion | 100 | Baseline |
| During the promotion | 150 | 50% higher |
| From September 14, 2026 | 125 | 25% higher |
The fall from 150 to 125 removes 25 units. To calculate the percentage reduction from the amount users had during the promotion, divide by 150:
(150 − 125) ÷ 150 × 100 = 16.7%
Calling this a “25% cut” would use the wrong denominator. The boost fell by 25 percentage points, while the total promotional allowance fell by one-sixth.
There is another useful way to see the effect. Imagine a fixed workload consumes 75 of those illustrative units. During the promotion, that would use 50% of a 150-unit week. Against a 125-unit week, the same workload would use 60%.
That is a calculation, not a measurement of Claude’s live usage meter. It shows why a smaller allowance can make the same amount of work occupy a larger share of the week without any change in the work itself.
Weekly limits and five-hour limits answer different questions
A five-hour session limit constrains usage within a shorter window. A weekly limit constrains usage across the longer cycle. Having room left in one does not guarantee room in the other.
Anthropic’s usage guidance says paid users can check both under Settings → Usage. The current-session indicator shows session consumption and time remaining; the weekly indicators show the relevant weekly limits and reset information.
If the weekly allowance is exhausted, waiting for a five-hour window to reset does not itself replenish that weekly allowance. If the session limit is the obstacle, the September change to the weekly promotion does not explain that session cap.
Read the specific limit message before deciding what changed. The distinction matters when you are trying to finish a job tonight rather than plan the rest of your week.
Why the same number of prompts may use a different amount
A prompt count is a poor measure of coding capacity. “Fix this typo” and “trace this failure through the repository, implement a fix, and verify it” can demand very different amounts of work.
Anthropic lists conversation length, attachment size, tool use, model choice, effort level, and multi-step tasks among the factors affecting consumption in its usage best practices. A smaller weekly allowance is therefore one possible explanation for reaching a limit sooner, not proof that every interruption has the same cause.
There is also a separate distinction between usage limits and context length. The first governs how much you can use the service over time; the second governs how much information fits in a conversation. Anthropic explains this in its guide to usage and length limits. Starting a fresh conversation should not be treated as a weekly-quota reset.
What to check before changing your plan
- Identify the exhausted limit. Check the session and weekly indicators and note the reset time shown for your account.
- Compare similar work. Record the model, effort setting, task scope, and usage before and after a typical job. Comparing a brief edit with an extended debugging session will tell you little.
- Check your billing route. Anthropic’s Pro and Max setup guide warns that an
ANTHROPIC_API_KEYenvironment variable can cause Claude Code to use API billing instead of subscription usage. Moving to API access changes how you pay.
For the next comparable coding task, write down the weekly percentage before you start and after you finish. Together with the account’s reset time, that gives you a practical basis for planning the remaining work.
Trending on Kingy
Keep reading with the stories getting the most attention now.