When does the Claude Code weekly limit promotion end?

The promotion ends August 19, 2026 at 11:59 PM PT. It ran from May 13, 2026 and increased weekly usage limits in Claude Code by 50% for Pro, Max, and Team plans plus legacy seat-based Enterprise users. After it ends, weekly usage limits return to standard levels with no change to your plan or billing.

Does the end of the promotion affect my 5-hour Claude Code limit?

No. The promotion only ever increased weekly usage limits. Five-hour session limits were not affected by the promotion and are unchanged when it ends. If you regularly hit the 5-hour cap rather than the weekly cap, this change does not affect you at all.

How do I check my Claude Code usage limits?

Run /usage in the Claude Code CLI. It shows your current weekly and 5-hour limits and how much you have consumed. Check it before August 19, 2026 to see your boosted ceiling, then again afterwards to see your standard one — the difference is exactly the third of your weekly headroom that is going away.

Quick Answer

Claude Code Weekly Limit Promo Ends August 19, 2026

Published: August 17, 2026

The Short Answer

Anthropic’s 50% weekly usage boost for Claude Code ends August 19, 2026 at 11:59 PM PT. It has been running since May 13, 2026 — over three months, long enough that most heavy users have quietly normalised it as their baseline.

After it ends, weekly limits return to standard levels. No plan change, no billing change. You simply get less.

What Exactly Changes

	During promo	After Aug 19
Weekly usage limit	+50%	Standard
5-hour session limit	Unchanged	Unchanged
Plan / billing	Unchanged	Unchanged

Who was included: Pro, Max, and Team plans, plus legacy seat-based Enterprise users.

Who was not: Free plans and consumption-based Enterprise seats. If you’re on a consumption-based Enterprise seat, nothing changes for you on August 19 — you never had the boost.

Where it applied: Claude Code only — CLI, IDE extensions, desktop, and web. Usage limits for Claude (web, desktop, mobile) and Claude Cowork were never part of this promotion and are unaffected.

The Maths Nobody Does

A 50% boost removed is not a 50% cut. If your boosted weekly limit was 150 units, standard is 100 — you lose 33% of the ceiling you have been working against.

Practically: if you have been finishing your week comfortably at around 90% of the boosted limit, you will hit the standard limit around Thursday.

Check Before It Goes

Run this in the Claude Code CLI:

/usage

It reports your current weekly and 5-hour limits and consumption against both. Do it now, while the boost is live, and note the weekly number. Do it again on August 20. The difference is your real exposure — not a guess from a forum post.

The distinction that matters most: are you a weekly-limit user or a 5-hour-limit user? If you routinely hit the 5-hour session cap and rarely approach the weekly one, August 19 is a non-event for you. If you have been running long unattended agent sessions across the week, it is not.

How To Absorb It

1. Route models by task difficulty. Not everything needs the top model. Reserve the expensive tier for architecture, debugging, and refactors that span files; push boilerplate, test scaffolding, and mechanical edits down a tier. This is the single biggest lever and it costs you almost no quality. See how to choose an AI model in 2026.

2. Stop re-reading the codebase. A large share of consumed context in agent sessions is the agent re-discovering files it already saw. Keep a project context file, be explicit about which paths matter, and scope requests to directories.

3. Compact aggressively. Long conversations resend their full history on every turn. Summarise into a state note and start fresh sessions at natural boundaries rather than letting one thread run for hours.

4. Batch the trivial. Ten separate one-line requests cost far more than one request describing ten one-line changes, because each round trip re-sends context.

5. Consider API spillover. If you genuinely need more than the standard weekly limit, per-token API access has no weekly cap. Claude Sonnet 5 sits at $2/$10 per MTok — and that rate is now permanent, after Anthropic cancelled the September 1 increase. For overflow work rather than all work, this is often cheaper than upgrading a plan tier.

The Bigger Pattern

This is the third time in 2026 that Claude Code limits have moved: a squeeze in March, a doubling of 5-hour limits in May, and this weekly promotion running May through August. The community reaction to the last extension was blunt — the promotion is what makes the product usable at heavy volume, and users would prefer the boosted numbers simply became the standard.

Anthropic hasn’t said whether it will extend again or make it permanent. Plan for the promotion ending, because that is what the published terms say. If it gets extended, you have lost nothing by tightening your usage; the habits above reduce cost and latency regardless of what the ceiling is.

One useful signal: the promotion’s own support article is titled “May–August 2026 weekly limits promotion.” That naming suggests a defined window rather than a rolling extension.

What To Do This Week

Run /usage today and record your boosted weekly limit.
Finish anything limit-heavy before August 19 — large refactors, bulk migrations, big test-suite generation.
Set up model routing so the drop lands on cheap tokens, not your critical work.
Run /usage again on August 20 to see the real standard ceiling instead of estimating it.

Last verified: August 17, 2026, against Anthropic’s official support documentation.