Anthropic unveils new rate limits to curb Claude Code power users

In a significant move to manage surging demand and curb misuse, Anthropic has announced the implementation of new weekly rate limits for its highly popular AI coding tool, Claude Code. The company stated that these measures are designed to address instances of continuous, round-the-clock usage by power subscribers and to prevent violations of its usage policy, including account sharing and unauthorized resale of Claude Code access.

The new rate limits are set to take effect on August 28, 2025, impacting subscribers across Anthropic’s paid plans: the $20-per-month Pro plan, as well as the $100- and $200-per-month Max plans. Anthropic communicated these changes to its subscriber base via email and through a public post on X (formerly Twitter) on July 28, 2025, stating that the changes are estimated to affect less than 5% of their total subscriber base.

While existing usage limits that reset every five hours will remain in place, Anthropic is introducing two new weekly rate limits that will reset every seven days. One of these limits will apply to overall usage, while the other is specifically tailored for Claude Opus 4, Anthropic’s most advanced AI model. Max plan subscribers will have the option to purchase additional usage beyond the set rate limits at standard API rates, offering some flexibility for heavy users.

This announcement follows a period of silent adjustments to Claude Code’s rate limits by Anthropic just weeks prior. The company has acknowledged the challenges in serving the broad demand for its AI coding tool, which has experienced “unprecedented demand since launch.” The official Anthropic status page indicates that Claude Code has faced partial or major outages at least seven times in the past month, suggesting that intense, nonstop usage by some users may have contributed to service disruptions.

Amie Rotherham, a spokesperson for Anthropic, affirmed in an email to TechCrunch that “most users won’t notice a difference” and reiterated that the new limits will impact less than 5% of subscribers based on current usage patterns. The company estimates that most Pro users can expect between 40 to 80 hours of Sonnet 4 through Claude Code within their weekly rate limits. For subscribers to the $100-per-month Max plan, this translates to 140 to 280 hours of Sonnet 4 and 15 to 35 hours of Opus 4. The $200-per-month Max plan subscribers can anticipate 240 to 480 hours of Sonnet 4 and 24 to 40 hours of Opus 4.

However, usage may fluctuate based on factors such as codebase size. The shift in advertised usage metrics has raised questions, as Anthropic previously claimed the $200 Max plan offered 20 times more usage than the Pro plan. With the updated figures, it appears subscribers now receive approximately six times as many Claude Code hours as Pro users, though Anthropic has not clarified if the 20x figure still applies when measured in tokens or computational resources.

This situation highlights a broader challenge within the AI industry: computational resource constraints. Anthropic, like many other AI model providers, is reportedly very constrained in this regard, with companies racing to bring new AI data centers online to meet the immense demands of serving and training advanced AI models.

The move by Anthropic mirrors similar adjustments seen across other AI coding tool providers. In June, Anysphere, the company behind Cursor, altered its pricing strategy for its $20-per-month Pro plan to curb power user abuse, though they later apologized for unclear communication regarding these changes. Replit, another prominent AI coding tool provider, also implemented comparable pricing changes in June.

Anthropic has assured its Claude subscribers of its commitment to “supporting long-running use cases through other options in the future.” For the short term, the company asserts that these new rate limits are crucial for maintaining reliable service across its user base.