Many AI companies had started off with seemingly unlimited $200 per month plans, but some of them now seem to be pulling back on them.
Anthropic has said that it’s introducing weekly rate limits for its power users, even for those on the Claude Pro and Max plans. The company said that this was because some users were abusing the nature of these plans and using “tens of thousands of dollars” of compute on a $200/month plan.
“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,” Anthropic said on X. “Some of the biggest Claude Code fans are running it continuously in the background, 24/7. These uses are remarkable and we want to enable them. But a few outlying cases are very costly to support. For example, one user consumed tens of thousands in model usage on a $200 plan,” it added.
“In other cases, a small number of users are violating our usage policies by sharing and reselling accounts. This impacts capacity for all Claude users, and we’re taking appropriate action to stop it,” Anthropic said. “From August 28, we’ll introduce new weekly limits that’ll mitigate these problems while impacting as few customers as possible. We’ll also support Max plan users buying additional usage at standard API rates, if they choose to.”
X users largely seemed largely unhappy with the new rules. “This was a stupid move. Qwen 3 models, Kimi k2 and now ZAI have dropped and Anthropic thought that this was the best time to give stricter rate limits,” wrote a user. Others threatened to cancel their subscriptions over the change.
Anthropic also sent users a mail saying that under the new rules, Max users can expect 240-480 hours of Sonnet 4 use and 24-40 hours of Opus 4 use weekly. Users who hit the weekly limit will have an option to purchase more compute at standard API rates.
Claude’s models are especially popular among coders, who find their performance to be superior to those dished out by its competitors. Agents like Claude Code can run for long periods in the background, and users can spin up multiple instances of each agent. As such, power users of these models likely were likely using up massive amounts of compute, and have forced Anthropic to pull back on its unlimited plans. It remains to be seen how competitors react to this move, but it does seem that unlimited access to models — while compute is still relatively expensive — might not be sustainable in the long term for most model companies.