Anthropic is implementing new weekly rate limits for its Claude artificial intelligence service, specifically targeting high-usage subscribers of Claude Code, effective August 28.
These new limitations aim to mitigate excessive continuous usage of Claude Code, particularly by subscribers operating the AI coding tool “continuously in the background, 24/7.” Anthropic also indicated that the rate limits are intended to address violations of Claude’s usage policy, which include account sharing and the unauthorized reselling of access to Claude Code.
The new weekly rate limits will apply to all subscribers across Anthropic’s pricing tiers: the $20-per-month Pro plan, and the $100- and $200-per-month Max plans. Anthropic communicated this development via an email to its subscribers and an official post on X.
We’re rolling out new weekly rate limits for Claude Pro and Max in late August. We estimate they’ll apply to less than 5% of subscribers based on current usage. pic.twitter.com/X8FAss3jIC
— Anthropic (@AnthropicAI) July 28, 2025
Anthropic confirmed that its existing usage limits, which currently reset every five hours, will remain active. In addition to these, two distinct weekly rate limits will be introduced, resetting every seven days. One of these new limits will be a general usage cap, while the second will specifically govern usage of Claude Opus 4, Anthropic’s most advanced AI model. Max subscribers will retain the option to acquire additional usage beyond their allocated rate limits by purchasing it at standard API rates.
This announcement follows a prior, unpublicized introduction of rate limits for Claude Code by Anthropic weeks earlier. At that time, Anthropic acknowledged existing issues but did not provide further details. While Claude Code has achieved significant adoption among developers, Anthropic has encountered challenges in scaling its service broadly. Anthropic’s official status page indicates that Claude Code has experienced either partial or major outages on at least seven occasions within the last month. This service disruption has been attributed, in part, to consistent, high-intensity usage from some power users.
Amie Rotherham, an Anthropic spokesperson, stated in an email to TechCrunch regarding the weekly rate limits, “Claude Code has experienced unprecedented demand since launch.” Rotherham further clarified that the new limits are expected to affect “less than 5% of subscribers,” based on their current usage patterns, and that “most users won’t notice a difference.”
Anthropic provided specific usage hour estimates for its subscribers under the new weekly limits. Most Pro users can expect to receive between 40 and 80 hours of Sonnet 4 through Claude Code. Subscribers to the $100-per-month Max plan are projected to receive 140 to 280 hours of Sonnet 4, alongside 15 to 35 hours of Opus 4. For subscribers on the $200-per-month Max plan, the expected usage includes 240 to 480 hours of Sonnet 4 and 24 to 40 hours of Opus 4.
Claude’s new feature lets users design AI tools and run them instantly
The company noted that actual usage may fluctuate depending on factors such as codebase size. Anthropic previously asserted that its $200 Max plan offered 20 times more usage than the Pro plan. However, based on the updated figures provided, the $200 Max plan now delivers approximately six times the Claude Code hours compared to the Pro plan. Anthropic has not clarified if the 20x figure remains applicable when measured in tokens or computational resources.
Anthropic has previously indicated that it faces significant constraints regarding computational resources, a challenge common among many AI model providers. Numerous AI companies are actively engaged in establishing new AI data centers to address the substantial demands associated with serving and training their AI models.
Other providers of AI coding tools have also revised their pricing strategies. In June, Anysphere, the company behind Cursor, modified its pricing for the $20-per-month Pro plan to restrict excessive usage by power users. Anysphere subsequently issued an apology for its inadequate communication of these changes, which resulted in some users incurring higher-than-anticipated costs. Replit, another provider of AI coding tools, implemented similar pricing adjustments in June.
In an email to Claude subscribers, Anthropic affirmed its dedication to “supporting long-running use cases through other options in the future.” The company maintains that these current rate limits are necessary to help ensure reliable service for the wider user base in the immediate term.
