Breaking

Anthropic Resets Claude Five-Hour and Weekly Limits as Claude Fable 5.1 Launches

September 1, 2026 at 19:58 EDT

Foundation Models
Software Dev & Coding

Anthropic has reset five-hour session limits and weekly usage caps for all Claude users as it released Claude Fable 5.1, restoring depleted counters on the same day the new frontier model became generally available. The flush is a one-time clear of existing meters, not a permanent increase in how much each plan includes.

September 1, 2026 · Anthropic

Usage caps reset as Fable 5.1 goes live

Anthropic flushed five-hour session limits and weekly usage caps for all Claude users on the day it shipped Claude Fable 5.1. Counters start over; plan rules do not.

All

Users had 5-hour and weekly counters reset

Rolling session window, still the daily throttle

50%

Max/premium weekly pool Fable may use

75%

Cheaper cache reads on token-billed usage

Cache reads now cost a quarter as much

Fable 5.1 cache reads are $0.25 per million tokens — 75% less than before. Column height is the relative price.

4×

Prior cache cost

1×

Fable 5.1 · $0.25/MTok

Takeaway: Long agent jobs that reread cached context get most of the savings.

Token-billed cost cut vs Fable 5

Each block is 5 percentage points of estimated savings. Same unit, drawn to scale.

~25%

Typical workloads

~45%

Highly agentic jobs

What users welcomed

A full quota refresh let people try Fable 5.1 immediately instead of waiting out a depleted week.

What still bites

Some Max users said Fable 5.1 can drain a five-hour session in well under an hour. Pro still pays credits.

The reset restores counters. It does not change the rules.

Max / premium

Fable still capped at 50% of the weekly pool

Pro / standard

Fable 5.1 stays on usage credits, with no new promo credit

After a Fable cap

Pay API rates on credits, or switch models and stay in-plan

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.