AI Pulse · Frontier Labs Watch

Anthropic says it will permanently raise Claude Code’s weekly limits by 25% from its normal baseline, effective September 14. On the surface, that sounds like more capacity for paying users.

The catch is timing. Since May, Claude Code subscribers have been receiving a temporary 50% boost. When that ends in September, the new permanent limit will sit at 125% of the original baseline, down from the 150% users can access today. In practical terms, that is a 17% reduction in weekly usage from current levels. Anthropic confirmed that framing after users pushed back on the announcement.

For a business using Claude Code in real workflows, this is a reminder to plan around actual available capacity, not headline percentages. If your developers, analysts, or operations team have built a weekly rhythm around the current allowance, they may hit limits sooner after September 14. That can mean interrupted work, more careful allocation across staff, or a review of which tasks genuinely need the higher-capacity model.

This is also the kind of usage change that should be visible in an AI command centre. Track which teams depend on each tool, how often they hit limits, and what the cost or productivity impact is when a provider changes its terms. AI vendors will keep adjusting pricing, access, and usage caps. The companies that notice early will have more options than those who find out when work stops.