Claude Code limits drop 17% on Sept 14 — what agencies should do

Published August 30, 2026Updated September 14, 2026By ABD Legacy LLC
Claude Code AI coding agents Agency operations Cost planning

The short version: Anthropic's temporary +50% Claude Code weekly-limit boost ended September 13, 2026 at 11:59 PM PT; from September 14, 2026 standard weekly limits are permanently 25% higher than the pre-promotion baseline for Pro, Max, Team, and seat-based Enterprise plans — which is about 17% below the promotional level, Anthropic's own figure. Normalize the pre-promotion baseline to 100 and the sequence is 100 → 150 (promotion) → 125 (permanent), so 125 ÷ 100 = +25% and 125 ÷ 150 = −16.7%. For agencies running Claude Code, the net effect versus the promotion is a 17% reduction in weekly capacity — the company confirmed the number itself: "Compared to today, this works out to a 17% reduction in weekly limits on Claude Code." If your delivery pipeline ran near the cap you had in August, your throughput ceiling is lower today and the cost-per-task math needs a re-run.

What changed: the +50% boost expired and a +25% baseline took over

Anthropic announced on X (via @ClaudeDevs, August 29, 2026) that it was permanently raising Claude Code's standard weekly usage limits by 25% for Pro, Max, Team, and seat-based Enterprise plans. The catch: the temporary 50% weekly boost — live since May 13, 2026 and extended through the summer — ended first, at 11:59 PM PT on September 13, 2026, one day before the new level took effect. Both dates now sit on Anthropic's own surface: its Help Center article "Claude Code May–August 2026 weekly limits promotion" carries a 2026-09-14 update stamp, states that the promotion "ran from May 13 through September 13, 2026", and states that the permanent level took effect "starting September 14, 2026" — see the update note below for the verbatim text.

"Starting September 14, we're permanently raising standard weekly limits in Claude Code by 25% for Pro, Max, Team, and seat-based Enterprise plans. Until then, the current 50% increase will be in place." — Anthropic via @ClaudeDevs, August 29, 2026 (the pre-change announcement, quoted verbatim)

Because the permanent level replaced the larger temporary boost, paid users lost capacity. The canonical illustration, used by BleepingComputer, MacObserver, and others:

PeriodWeekly limit (relative to pre-promotion baseline)
Pre-promotion baseline (before May 13, 2026)100%
Promotion — May 13 through Sept 13, 2026 (temporary +50% boost, now ended)150%
From Sept 14, 2026 (permanent — the level in force today)125%

Going from 150 to 125 is a 16.7% cut — 25 ÷ 150 = 16.7%, which Anthropic and every major outlet round to about 17%. Read against the old baseline instead and the same 125 is a 25% increase (125 ÷ 100 = 1.25): both headlines describe one limit and differ only in the denominator. Returning to the promotional level would take a further +20% (25 ÷ 125).

Why it matters: agency throughput and cost-per-task

Claude Code is a core delivery tool for many agencies: automations, client sites, internal tooling. A weekly limit is not a soft suggestion — it is the ceiling on how much agent work your seats can produce in a week. Two effects follow:

What did not change: Anthropic's promotion page states the boost "did not affect" five-hour session limits, and no five-hour change has been announced since; the permanent sentence covers weekly limits only, and no plan-price or API-billing change was announced. This is a weekly-capacity change only. The permanent sentence names Pro, Max, Team, and seat-based Enterprise plans and does not restate the promotion's exclusions, so no claim is made here about Free plans or consumption-based Enterprise seats either way.

How many Claude Code requests per week do you get?

The honest answer: Anthropic does not publish weekly Claude Code limits as fixed message or token counts. Usage varies by model, conversation length, tool usage, and effort level, which is why the company communicates limits as ratios rather than numbers. What you can verify for your own account: on Pro, Max, Team, and seat-based Enterprise, open Settings > Usage to see progress bars for your five-hour and weekly limits. That is the only reliable way to know how close you are running to the ceiling.

Claude Code usage limits 2026: what changed and what didn't

Update — September 14, 2026. The documentation gap is closed. Anthropic's Help Center article "Claude Code May–August 2026 weekly limits promotion" (support.claude.com/en/articles/15910845) now carries a 2026-09-14 update stamp and states verbatim that the promotion "ran from May 13 through September 13, 2026", that the offer was "valid from May 13, 2026 through September 13, 2026 at 11:59 PM PT", and that "starting September 14, 2026, weekly limits in Claude Code are 25% higher than they were before the promotion for Pro, Max, Team, and seat-based Enterprise plans". This page's earlier version was written on August 30, 2026, when the Help Center still listed an August 31 end date and the September 14 schedule came only from Anthropic's @ClaudeDevs posts. That forward-looking copy is replaced by this refresh.

Claude Pro vs Max limits: what we know

Agencies choosing between Claude Pro and Claude Max often ask how the weekly limits compare. What is verified: both plans are covered by the September 14 change, and each shares a single weekly pool between the Claude apps and Claude Code. What is not published: fixed weekly limit numbers for either plan. Anthropic has never published the absolute values, so treat any specific "requests per week" figure you see elsewhere as an estimate, not an official number. The verified comparison is the ratio — the promotional 150% of baseline, replaced by 125% from September 14, 2026 for both tiers.

5 actions to take now the change has landed

  1. Re-run your budget on the standard cap. Model capacity at 125/150 (0.833×) of the promotional limit. If a workload used 100% of the promotional cap, it is 120% of the standard cap today — decide whether you absorb it, raise retainers, or add headroom. The worked model — seats, credits and the over-cap case — is on Claude Code usage limit cost impact at aiagencycalculator.com.
  2. Rebalance against the new ceiling. The boost ran through September 13, 2026 and is gone. Any spike work still planned against the promotional ceiling needs a new slot — stagger big refactors, batch migrations and client build-outs, or schedule them as usage-credit work.
  3. Review your plan mix and seat count. If you are near the cap weekly, the post-Sept-14 math may justify upgrading seats or adding capacity. Check Settings > Usage to see your real weekly consumption before you decide — do not guess.
  4. Add efficiency tooling. Every message you save is capacity you keep. CLAUDE.md memory files, compact prompts, well-scoped subagents, and caching cut messages-per-task and stretch the weekly bucket further.
  5. Set a review date. The change landed September 14, 2026 — compare usage from before and after it, then adjust client pricing or internal estimates on the new cost-per-task.

None of this requires panic. The change is a capacity re-baseline, not a service interruption — but it landed at budget season for many agencies, which is why the numbers are worth re-running now rather than at the next renewal.

Frequently asked questions

How many Claude Code requests per week do you get?

Anthropic does not publish weekly Claude Code limits as fixed message or token counts — usage varies by model, conversation length, tool usage, and effort level. The September 14, 2026 change is a ratio: limits ran at 150% of the pre-promotion baseline during the promotion and are a permanent 125% since, a 16.7% (about 17%) cut versus the promotional level.

What are the Claude Code usage limits in 2026?

For Pro, Max, Team, and seat-based Enterprise plans, Claude Code weekly usage limits ran 50% above the standard baseline from May 13 through September 13, 2026. Since September 14, 2026 they are a permanent 25% above the pre-promotion baseline — a net 17% reduction versus the promotion. The promotion did not affect five-hour session limits, and no plan-price or API-billing change was announced.

What are the Claude Pro vs Max limits?

Both Claude Pro and Claude Max are covered by the change, which took effect September 14, 2026. Pro and Max each share a single weekly usage pool across the Claude apps and Claude Code. Anthropic has not published fixed weekly limit numbers for either plan, so the 100-150-125 ratio is the only verified comparison.

When did the Claude Code weekly limit change take effect?

The temporary 50% boost ran from May 13, 2026 and ended September 13, 2026 at 11:59 PM PT. The permanent +25% baseline took effect September 14, 2026, per Anthropic's announcement on X (ClaudeDevs) and the vendor's Help Center article, which now documents the change verbatim.

Is the Claude Code 50% boost still active?

No. The temporary 50% weekly-limit boost launched May 13, 2026 and ended September 13, 2026 at 11:59 PM PT. Since September 14, 2026 standard weekly limits are permanently 25% higher than the pre-promotion baseline for Pro, Max, Team, and seat-based Enterprise plans — about 17% below the promotional level.

Planning your agency's coding-agent stack? Compare the full pricing picture in our AI coding agent pricing guide — billing models, cost variables, and how to budget for Claude Code, Codex, Cursor, and the rest.

See the AI coding agent pricing guide →

Sources

Accuracy note: refreshed 2026-09-14 against the primary source — Anthropic Help Center, "Claude Code May–August 2026 weekly limits promotion" (support.claude.com/en/articles/15910845), whose update stamp reads 2026-09-14 — plus Anthropic's @ClaudeDevs posts of August 29, 2026. The earlier version of this page was verified 2026-08-30; that forward-looking copy is superseded. Anthropic publishes no absolute weekly limit values for any plan, so every figure here is a ratio (100 → 150 → 125). No grandfathering or regional exceptions were reported in any fetched source, and uncorroborated third-party plan multipliers were intentionally excluded.