Anthropic says its rolling out new weekly rate limits for Claude to curb usage among subscribers who are running its AI coding tool, Claude Code, “continuously in the background, 24/7.” Anthropic says the rate limits also aim to stop a handful of users who are violating Claude’s usage policy by sharing accounts and reselling access to Claude Code.
The new rate limits will go into effect August 28 for subscribers to Anthropic’s $20-per-month Pro plan, as well as its $100 and $200-per-month Max plans, the company said Monday in an email to subscribers and a post on X.
Anthropic says its existing usage limits, which reset every five hours, will remain in place. The company is also introducing two new weekly rate limits that reset every seven days; one is an overall usage limit, whereas the other is specific to Anthropic’s most advanced AI model, Claude Opus 4. Anthropic says Max subscribers can purchase additional usage, beyond what the rate limit provides, at standard API rates.
The announcement comes just weeks after Anthropic quietly introduced rate limits for Claude Code. The company said at the time it was aware of the issues, but declined to elaborate further. While Anthropic’s AI coding tool has a been a hit with developers, the company seems to be having a difficult time serving it broadly. Anthropic’s status page shows that Claude Code has experienced a partial or major outage at least seven times in the last month — perhaps because some power users seem to be running Claude Code nonstop.
“Claude Code has experienced unprecedented demand since launch,” said Anthropic spokesperson Amie Rotherham in an email to TechCrunch about the weekly rate limits. Rotherham notes that “most users won’t notice a difference,” however, did not offer specifics about how much usage subscribers should now expect.
The company notes that this limit will affect less than 5% of subscribers, based on their current usage patterns. Most Pro users can expect 40-80 hours of Sonnet 4 within their weekly rate limits, according to Anthropic. However, Anthropic notes that usage may vary based on codebase size and other factors.
Anthropic has said before that it’s very constrained when it comes to computational resources, which seems to be the case for most AI model providers today. Most AI companies are racing to bring new AI data centers online to meet the massive demands of serving and training their AI models.
Techcrunch event
San Francisco
|
October 27-29, 2025
Several providers of AI coding tools are revisiting the pricing strategy around their products. In June, the company behind Cursor, Anysphere, changed the way it priced usage for its $20-per-month Pro plan to limit power users from abusing the plan. However, Anysphere later apologized for poorly communicating those changes, leading to some users paying more than they expected. Another AI coding tool provider, Replit, made similar pricing changes in June as well.
In an email to Claude subscribers, Anthropic says it’s committed to “supporting long-running use cases through other options in the future.” However, the company claims these rate limits will help them maintain reliable service broadly in the short term.
Update 1:37 pm PT: This story was updated to include comments from an Anthropic spokesperson.
