Claude Code Weekly Limits Fall 17% Today as Anthropic's…

Anthropic's four-month promotional boost on Claude Code weekly limits expired at 11:59 PM PT on September 13, 2026, and as of yesterday every Pro, Max, Team, and legacy seat-based Enterprise account is running against a ceiling 17% below what it had the day before. The company's framing calls this a permanent raise. The arithmetic calls it a cut.

Both are true. Index the pre-promotion baseline at 100. The May 13 promotion pushed the weekly ceiling to 150. As of September 14 it sits at 125, which is indeed 25% above where things started but 125 ÷ 150 = 0.833, and recovering a 17% cut takes a 20% gain. Anthropic's own developer account, per BleepingComputer, described it plainly: "a 17% reduction in weekly limits on Claude Code."

The Help Center article, rewritten yesterday, threads the same needle: "This promotion was a limited-time boost on top of your plan. It ran from May 13 through September 13, 2026… 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." Free tiers and consumption-based Enterprise seats were never in the promotion and aren't affected now. Five-hour rolling limits didn't move either.

The wrinkle is that nobody outside Anthropic knows what any of these ceilings are in absolute tokens. Digital Applied and AI Catchup both note Anthropic has never published the numbers, before, during, or after. The only reading with a user's name on it comes from /usage in the CLI, or from the status-line JSON keys (rate_limits.seven_day, rate_limits.seven_day_opus, rate_limits.seven_day_sonnet) that Digital Applied documents against build 2.1.237. Weekly allowance doesn't map to prompt count, Anthropic told BleepingComputer, because usage varies with conversation length, model choice, tool selection, and effort level.

That's the awkward part for anyone who calibrated an agentic pipeline to the 1.5x promotional ceiling over the summer. The lost headroom is roughly a sixth, and it hits the exact workloads that made Claude Code interesting in the first place: context-heavy refactors, multi-day subagent fan-outs, spike weeks, experimentation runs. Pipelines that ran comfortably through Friday may now stall Thursday afternoon.

Anthropic has said on X, per The Decoder, that further changes are coming that'll "make it feel like you're getting more from Claude," with more visibility and control. The verb choice does the work. Feeling and getting aren't the same thing, a distinction the pricing page has now demonstrated twice: Fable 5.1's business-automation benchmark gains and Real SWE frontier-agent results are the capability story; weekly limits are the settlement.

Claude Code Weekly Limits Fall 17% Today as Anthropic's Summer Promo Expires

Sources