hunter-gplpcn
Karl Kahn paid $200 a month for Anthropic’s most expensive Claude subscription. Within weeks, he says, the arithmetic stopped working. A single five-hour coding session consumed 15% of his weekly usage allowance, according to the complaint, forcing him to stop mid-task or ration a product he had paid to use at maximum capacity.
Kahn, a Washington, D.C. resident, filed a federal lawsuit on Monday accusing Anthropic of false advertising over the usage limits attached to its two highest-priced plans, Max 5x and Max 20x. The first costs $100 a month; the second costs $200. The suit seeks class-action status on behalf of everyone who bought either plan since April of last year, according to The Wall Street Journal, which first reported the case.
The complaint rests on a simple allegation: Anthropic marketed the plans as five times and twenty times the capacity of its base tier, but the weekly caps make that promise unreachable in practice for the heavy users the plans are designed to serve. Kahn says he upgraded to Max 20x expecting the top tier to end his usage worries, and instead found himself checking a counter that ran down faster than the marketing suggested.
A technical detail features prominently in the suit. Anthropic’s usage limits reset on rolling five-hour windows rather than on a clean daily or monthly schedule, according to the complaint, a design that makes it difficult for subscribers to predict when their capacity replenishes and hard to plan work around it. For a developer in the middle of a long coding session, the result is a quota that evaporates invisibly, the complaint argues.
Anthropic has not publicly commented on the lawsuit. The company’s plans are among the most expensive in the consumer AI market, and its marketing has emphasized the productivity gains that heavy users can unlock. Whether the multipliers advertised on the product pages match what subscribers actually receive is now a question for the courts.
The case arrives at a delicate moment for the AI subscription market. OpenAI and Anthropic are both tightening usage policies and raising prices, according to a Business Insider report, as the industry moves from generous access to metered consumption. Consumers are spending real money on AI subscriptions, and complaints about opaque usage accounting have become a fixture of app-store reviews and developer forums. This lawsuit is among the first to test those complaints in federal court.
On forums where Claude users compare notes, the Max plans have been a running subject: how much does a “20x” plan actually buy? Users post screenshots of quota meters and trade strategies for stretching allowances across a working week. The lawsuit gives that informal grumbling a formal shape, and the plaintiffs’ lawyers will be able to draw on the paper trail users have left behind.
The stakes extend beyond Anthropic. Consumer-protection lawyers said the case will hinge on what a reasonable user would understand from the marketing, a standard that has been litigated for decades in telecom, software and cloud services. Class certification is a hurdle, discovery could force Anthropic to open its usage-metering systems, and a settlement is possible at any point. But the precedent matters: if the case survives, every AI company selling tiered subscriptions will review its language about limits.
Claude’s position makes the case commercially significant. Anthropic converts about 13% of its monthly active users into paying subscribers, the best rate in the industry, according to Sensor Tower’s State of AI report. Subscription revenue is becoming a material part of Anthropic’s business, and a lawsuit that chills high-tier sign-ups would land where the company can least afford it.
Legal analysts said the case is an early example of consumers using class actions to push back against AI subscription costs and transparency, a front that is moving faster than regulation. The legal theory is familiar; the product category is new. People pay hundreds of dollars a month for AI tools they increasingly depend on for work, and they expect the limits to be legible.
The suit also names the specific mechanics of Kahn’s frustration: after upgrading, he found the weekly allowance inadequate for the coding work the plan’s marketing targets, a gap between promise and delivery that the complaint says is the essence of the claim.
The case is early. Courts move slowly, and a class-action suit of this kind typically takes years to resolve. But the complaint lands in a market where trust in usage accounting is thin, and where a precedent could reshape how AI companies describe what they sell. For now, Kahn’s lawsuit describes a simple mismatch: a $200 promise and a weekly counter that ran out. Whether that mismatch amounts to fraud is a question a federal judge will now answer.
