AI agents · OpenClaw · self-hosting · automation

Quick Answer

Claude Code +50% Weekly Limits Extended to Aug 31

Published:

The Short Answer

The 50% Claude Code weekly limit increase runs through August 31, 2026. Anthropic announced the extension on August 18-19, 2026 via its @ClaudeDevs account and said it hopes to make the boost permanent.

Detail
Boost size+50% weekly Claude Code limits
Current end dateAugust 31, 2026
Applies toPro, Max, Team, seat-based Enterprise
Permanent?Anthropic “hopes to” — not committed
Extension historyRunning since ~May 2026: Jul 13 → Jul 19 → Aug 19 → Aug 31

Verified August 20, 2026.

What Was Announced

The promotion is not new. Anthropic has been running a +50% weekly limit increase on Claude Code since roughly May 2026, extending it in short hops rather than committing to it — July 13, then July 19, then August 19, and now August 31, 2026.

What changed in the August announcement is the framing. Anthropic said it hopes to make the increase permanent, citing strong demand and additional compute secured. That is the first time the promotion has been discussed as a candidate for becoming the baseline rather than a rolling giveaway.

It covers Pro, Max, Team and seat-based Enterprise plans.

Why This Keeps Getting Extended

Weekly limits on Claude Code exist because agentic coding consumes tokens at a rate that ordinary chat does not. A single long autonomous session can burn through what a month of chat would. Anthropic introduced weekly caps to stop a small number of extremely heavy users from consuming a disproportionate share of capacity.

The repeated extensions tell you two things. First, the standard limits bind harder than Anthropic would like — enough users hit them that removing the boost is visibly unpopular. Second, capacity is the constraint, not policy. Anthropic tied this extension to having secured additional compute, which is the honest version of why limits move.

What To Actually Plan For

Budget as if the boost ends on September 1. That is the disciplined position. The promotion has been extended four times, which makes a fifth extension likely — but “likely” is not a capacity plan. If your team’s throughput depends on the extra 50%, you are exposed to a policy decision you do not control.

If your weekly headroom suddenly tightens in early September, this is why. It will not be a bug or a silent downgrade. It will be the promotion lapsing.

Three ways to reduce exposure:

  1. Route non-frontier work off Claude Code. Not every task in an agent session needs Opus 5. Cheaper models handle test generation, dependency bumps, docs and mechanical refactors at a fraction of the cost — Gemini 3.7 Flash at $0.75/$3.75 or GLM-5.3 at $1.40/$4.40 per MTok versus Opus 5 at $5/$25.

  2. Use the API for overflow rather than upgrading plans. Claude Code supports API-key billing. When the weekly subscription allowance runs out, metered API access continues without a plan change, and you can cap it.

  3. Watch what Fable 5 costs you. Fable 5 remains restricted to Max and premium Team and seat-based Enterprise seats, is capped at 50% of the weekly allowance, and weighs roughly double an Opus session against it. If you are hitting weekly limits and using Fable habitually, that is usually the single largest lever.

The Pricing Context

Separately and more durably: Claude Sonnet 5 pricing at $2/$10 per MTok is now permanent. The increase to $3/$15 that had been scheduled for September 1, 2026 was cancelled, per Anthropic’s documentation around August 10, 2026.

That is worth keeping distinct from the limits story. Two different September 1 events were circulating in August 2026 — a price rise that will not happen, and a limit boost that may or may not be extended. The price is settled; the limits are not.

Opus 5, launched July 24, 2026, remains at $5/$25 with a 1M context window and 128K max output — the same price as Opus 4.8 and unchanged since Opus 4.5 in November 2025.

Last verified: August 20, 2026.

Sources