Anthropic Unveils New Rate Limits To Curb Claude Code Power Users

Anthropic says its rolling retired caller play complaint limits for Claude to curb usage among subscribers who are moving its AI coding tool, Claude Code, “continuously successful nan background, 24/7.” Anthropic says nan complaint limits besides purpose to extremity a fistful of users who are violating Claude’s usage argumentation by sharing accounts and reselling entree to Claude Code.

The caller complaint limits will spell into effect August 28 for subscribers to Anthropic’s $20-per-month Pro plan, arsenic good arsenic its $100 and $200-per-month Max plans, nan institution said Monday successful an email to subscribers and a post connected X.

Anthropic says its existing usage limits, which reset each 5 hours, will stay successful place. The institution is besides introducing 2 caller play complaint limits that reset each 7 days; 1 is an wide usage limit, whereas nan different is circumstantial to Anthropic’s astir precocious AI model, Claude Opus 4. Anthropic says Max subscribers tin acquisition further usage, beyond what nan complaint limit provides, astatine modular API rates.

The announcement comes conscionable weeks aft Anthropic softly introduced complaint limits for Claude Code. The institution said astatine nan clip it was alert of nan issues, but declined to elaborate further. While Anthropic’s AI coding instrumentality has a been a deed pinch developers, nan institution seems to beryllium having a difficult clip serving it broadly. Anthropic’s status page shows that Claude Code has knowledgeable a partial aliases awesome outage astatine slightest 7 times successful nan past period — possibly because immoderate powerfulness users look to beryllium moving Claude Code nonstop.

Anthropic has said earlier that it’s very constrained erstwhile it comes to computational resources, which seems to beryllium nan lawsuit for astir AI exemplary providers today. Most AI companies are racing to bring caller AI information centers online to meet nan monolithic demands of serving and training their AI models.

Several providers of AI coding devices are revisiting nan pricing strategy astir their products. In June, nan institution down Cursor, Anysphere, changed nan measurement it priced usage for its $20-per-month Pro scheme to limit powerfulness users from abusing nan plan. However, Anysphere later apologized for poorly communicating those changes, starring to immoderate users paying much than they expected. Another AI coding instrumentality provider, Replit, made akin pricing changes successful June arsenic well.

Anthropic did instantly respond to TechCrunch’s mobility astir nan specifications of nan complaint limits.

The institution notes that this limit will impact little than 5% of subscribers, based connected their existent usage patterns. Most Pro users tin expect 40-80 hours of Sonnet 4 wrong their play complaint limits, according to Anthropic. However, Anthropic notes that usage whitethorn alteration based connected codebase size and different factors.

In an email to Claude subscribers, Anthropic says it’s committed to “supporting long-running usage cases done different options successful nan future.” However, nan institution claims these complaint limits will thief them support reliable work broadly successful nan short term.