(Credit: Anthropic)
Anthropic has introduced weekly rate limits for its Claude chatbot focused on curbing excessive—and forbidden—usage of Claude Code.
"Claude Code has experienced unprecedented demand since launch [on June 4]," the company says. "However, we've identified extreme usage by a small number of customers that impacts capacity for our broader community."
The rate limits take effect on Aug. 28 for those who subscribe to the $20-per-month Pro plan, $100 Max plan, and $200 Max plan.
The problem is that some subscribers are selling access to their login credentials to help cover the cost of premium plans. However, that violates Anthropic's policy and drives up usage on each account. Others "are turning Claude AI Pro account sharing into a small side hustle by renting access or offering AI-powered services based on it," says DICloak.com
Other overeager vibe coders are "multi-clauding," or running five instances of Claude Code in parallel. They kick off different tasks in each one and return once it's ready for review. Only an estimated 5% of "power users" are doing this, Anthropic tells us.
Multi-clauders draw down their rate limits faster, so they will be more likely to be impacted by the weekly limits. Anthropic says this is necessary "to keep the service reliable, and most users won't notice a difference."
Previous rate limits, which went into effect in July, prompted backlash from Claude Code users. As TechCrunch reports, subscribers hit those limits faster than usual and had to wait for them to reset, usually for a few hours.
It's easy to see how changing the terms of a subscription, especially one that's hundreds of dollars per month, would not go over well with users. But the computing power behind chatbots like Claude is wildly expensive and in short supply. Even heavyweights like OpenAI struggle to find enough GPUs for new launches. Anthropic wants power users to pay extra for their activity, and notes that "Max users can still purchase additional capacity at standard API rates when needed."