What’s going on?
Short version: Anthropic gave Claude Code a temporary power-up, then quietly turned part of it back into a permanent—yet smaller—upgrade. Confused? Welcome to corporate math.
Right now users are enjoying a temporary boost of 50% extra weekly usage. That extra juice lasts until September 13. Starting September 14, the company is switching to a new permanent limit that is 25% higher than the original baseline. Sounds generous… until you do the math.
Example time (because numbers don’t lie): if your original weekly allowance was 100 prompts, you currently have 150 thanks to the temporary boost. On September 14 that will drop to 125. So yes, you end up with 25% more than the pre-boost plan, but about a 17% reduction compared to what you have now.
What it means for you — and how to adapt
Here’s the practical bit: this is mostly a rebalancing move, not a big new gift. The company even pulled the initial post and then admitted that the permanent setup is lower than the temporary one. They also say other usage changes are coming, promising more visibility and controls.
- Who sees the change: Pro, Max, Team, and seat-based Enterprise plans—basically paid tiers.
- Why exact usage varies: your effective weekly allowance depends on conversation length, which model you pick, any tools used, and how much effort the AI spends on each prompt. So weekly quotas aren’t a fixed number of prompts.
- Timing: the temporary 50% increase stays until September 13. The new permanent limits take effect on September 14.
- What to do: track your usage this month, trim long-running threads, and consider switching models or workflows if you need to stretch the new allotment.
One last thing unrelated but worth remembering: when attackers get valid credentials, defensive tools often struggle — only about 37% of their actions get blocked in some large simulations. The bigger picture is that prevention drops after initial access, so keep your credential hygiene tight while you stare at your new limits.