
Claude Max Weekly Limits: The 50% Summer Boost Becomes a Permanent 25% Raise — Math Included
On September 14, 2026, Anthropic is replacing the temporary 50% weekly boost to Claude Code usage limits with a permanent 25% increase to standard baseline limits across Pro, Max, Team, and seat-based Enterprise plans. The headline reads as a raise. Against the meter that active users have actually been running on all summer, it is a cut of roughly 17%.
Yes, you read that correctly: during the summer the weekly Claude Code limit was boosted 50% on a limited-time basis. Now the boost is coming off — and what replaces it is a permanent 25% increase over the pre-promo baseline. Permanent is the genuinely good part: it is no longer a limited-time promotion, and unlike the summer boost it is not going to expire. But the arithmetic of going from a 50% boost to a 25% increase means most heavy users lose about a sixth of the weekly headroom they have today, and the announcement has landed with a wave of community backlash over what many read as a quietly disguised downgrade.
The Two Numbers, Side by Side
| Temporary 50% boost | Permanent 25% increase | |
|---|---|---|
| Effective period | Summer promo, active through September 13 | Starts September 14 |
| Plans covered | Claude Code weekly limits | Pro, Max, Team, and seat-based Enterprise |
| Duration | Limited-time | Permanent |
| Multiplier vs. baseline | 1.5x | 1.25x |
| Net vs. what users have today | — | ~17% less weekly headroom |
The distinction that matters is baseline vs. boost. The 25% is a permanent raise off the standard baseline limits. The 50% was a temporary multiplier on top of that same baseline. Because the boost and the raise never stack, the transition on September 14 is not additive — it is a swap, and the swap is downward.
The Timeline: Every Limit Change That Got Us Here
Claude's usage limits have moved at least five times since August 2025, and each change touched a different dial. This is the dated record:
- August 28, 2025 — Weekly rate limits are added on top of the existing rolling 5-hour window for Claude Pro and Max. Anthropic's stated reasons: a small number of users running Claude Code essentially 24/7, account sharing, and unauthorized reselling degrading availability for everyone else.
- March 13–28, 2026 — A temporary off-peak promo doubles 5-hour limits outside peak hours (8 AM–2 PM ET weekdays) and all weekend, for Free, Pro, Max, and Team. Bonus usage does not count against weekly caps. It ends on schedule; Enterprise is excluded.
- May 6, 2026 — Claude Code's 5-hour limits are permanently doubled and peak-hour throttling is removed for Pro and Max. This change is still in effect and is not part of the September adjustment.
- July 18, 2026 — Anthropic announces a temporary 50% weekly Claude Code boost, later extended through August 19 and again through September 13.
- September 14, 2026 — The 50% promo ends and the permanent 25% baseline increase takes effect for Pro, Max, Team, and seat-based Enterprise plans.
The weekly cap has been the one constant across all of it. Neither the March promo nor the May 6 doubling touched it — and it is exactly the weekly number that is about to change.
The Math: Why "25% Increase" Is a 17% Cut
The arithmetic is simple, which is probably why the backlash was so fast. Take a hypothetical 100-unit weekly baseline:
| Weekly baseline | With 50% boost (summer) | With 25% permanent (Sept 14+) |
|---|---|---|
| 100 units | 150 units | 125 units |
Going from 150 back down to 125 is a reduction of 25 units — which is 16.7% less than the boosted meter users have today. Or flipped around: anyone who is currently hitting the boosted cap will lose about one-sixth of their weekly capacity the day the permanent raise arrives.
Anthropic's framing is that the permanent 25% raise is structurally better than a temporary 50% promo: it never expires, it is predictable, and planning around a stable number beats planning around a countdown. That argument is real. It is also true that for the users the promo was designed to keep happy — the power users burning weekly limits every cycle — the immediate experience is a noticeable shrink, and the community reaction has been dominated by exactly that reading.
What Actually Stays Permanent
Two things worth calling out so nothing gets lost in the backlash:
- The 5-hour rolling window remains doubled. The May 6, 2026 change permanently doubled Claude Code's 5-hourlimits and removed peak-hour throttling for Pro and Max. That was never a promo and it is not being walked back on September 14.
- The weekly cap is the variable that is changing. Everything else about how limits meter is unchanged. The UI label "All models" in the usage panel is simply the aggregate weekly limit tracker — the cap still exists, and the September 14 change updates the number behind it.
So the honest summary: Claude Code's short-window capacity stays doubled, its weekly capacity settles at 25% above the original baseline, and for everyone who has been coasting on the summer boost, that combination is a step down from where they are today.
The Takeaway
The September 14 change is good news and bad news depending on where you sit. If you were below limits all summer, a permanent 25% raise off baseline is strictly better than any expiring promo — more headroom than pre-May, stable, and never going away. If you are a heavy weekly-cap-burner, the day the boost ends feels like a downgrade, because it is: ~17% less weekly capacity than the meter you have been running on.
The durable reads are these: Anthropic is signaling that limits are now a permanent, predictable feature of paid plans rather than a promotional toy, and the May doubling of the 5-hour window remains the substantive long-term gift to agent users. Budget your week on the new 1.25x baseline — and treat any "no limits" screenshot as a promo banner or a server hiccup, not a policy.
Sources: Anthropic ClaudeDevs announcement (X) · explainx.ai Claude usage limits timeline · daily.dev · Bleeping Computer · Android Headlines
Comments
Leave a comment
Related articles

Claude Fable 5.1: Same $10/$50 Price, 75% Cheaper Cache Reads, and Benchmarks That Double the Gap
Claude Fable 5.1 is live — same $10/$50 rate card as Fable 5, but cache reads drop 75% to $0.25/1M, cutting typical workload cost ~25% and agentic cost up to ~45%. Full benchmark table vs. Fable 5, Opus 5, and GPT-5.6 Sol.

Ox Alpha: The Anonymous 1M-Context Coding Model That Has Everyone Guessing Z.ai
Ox Alpha is a free, anonymous 1M-context reasoning model on OpenRouter. Community fingerprinting points to Zhipu AI's GLM-5.x. We break down the benchmarks, the mystery, and the adoption.

DeepSeek V4 Pro Is Now an Ollama Cloud Model: The 1.6T Flagship on a $20 Flat Plan
DeepSeek's 1.6T flagship V4 Pro is now a flat-rate Ollama Cloud model — $20 a month, one line of shell, drop-in for Claude Code and OpenCode. Here's what the 0813 refresh changed and when the plan beats the API.
