Anthropic's $200 Claude plan faces a lawsuit over what "20x" buys

Why it matters

The case tests whether AI companies can market simple usage multiples while separate weekly and model-specific caps sharply reduce what paying customers receive.

Anthropic has been fighting a proposed class action since June 14th over whether its $200-a-month Claude Max plan delivers the "20x" usage printed on the label.

The dispute resurfaced Monday when Ananth (@Ananth7e) circulated a table from the complaint comparing Anthropic's advertised subscription multiples with weekly limits the AI developer disclosed to customers. The filing makes a narrower claim than the post's assertion that the lawsuit "confirms" users' complaints: the figures remain allegations, the proposed class has not been certified and no court has ruled that Anthropic misled subscribers.

Karl Kahn, a Washington, D.C. Claude user, filed the 35-page complaint in the U.S. District Court for the Northern District of California. Kahn alleges that Anthropic's Max 5x and Max 20x plans provide substantially less weekly usage than their names suggest, while charging $100 and $200 a month, respectively.

The case puts a basic pricing question before the court: whether an AI subscription can advertise a large usage multiple for one time window while imposing separate caps that produce a much smaller multiple over a week.

The plaintiff's math

The complaint relies on usage estimates Anthropic sent subscribers in July 2025. According to the filing, Anthropic told Pro customers that most users could expect 40 to 80 hours of Sonnet 4 usage a week. Max 5x customers were told to expect 140 to 280 hours, while Max 20x customers were given a range of 240 to 480 hours.

Kahn's lawyers calculated that those ranges made Max 5x roughly 3.5 times Pro's weekly Sonnet allowance and Max 20x roughly six times Pro's, rather than five and 20 times. The complaint also notes that the $200 plan's stated Sonnet allowance was about 1.7 times the $100 plan's allowance, although its name implies a fourfold step from 5x to 20x.

For Opus 4, Anthropic's customer emails listed 15 to 35 hours a week for Max 5x and 24 to 40 hours for Max 20x, according to the filing. Anthropic did not provide a corresponding Pro range in those emails. The complaint estimated the Pro allowance by dividing the Max 5x range by five, then concluded that Max 20x provided approximately six to eight times Pro usage.

Those calculations are the plaintiff's analysis of Anthropic's figures, not an independent measurement of Claude's capacity. The emails also said actual use would vary with codebase size, parallel Claude Code instances and other settings. Their weekly-hour estimates can exceed the 168 hours in a calendar week because users can run multiple instances simultaneously.

Kahn says he subscribed to Claude Pro in June 2025, upgraded to Max 5x on January 5th and moved to Max 20x on April 21st. Within weeks, the complaint alleges, he encountered weekly caps and found that fully using one five-hour session could consume about 15% of his weekly allowance. He says he sometimes bought additional usage after reaching those limits.

The proposed class covers U.S. consumers who bought or upgraded to Max 5x or Max 20x through Claude's website or desktop application from April 9th, 2025 onward. The original complaint brought claims under California consumer protection and false advertising laws, along with negligent misrepresentation, breach of contract and unfair competition claims. It seeks restitution, damages, an injunction and class certification.

Anthropic defines "20x" by the session

Anthropic's current pricing page still presents Max as offering a choice of "5x or 20x more usage than Pro" near the top of the plan card. Farther down, Anthropic defines that multiple as usage per five-hour session and says paid subscriptions also carry weekly limits. The page states that Anthropic may impose weekly, monthly, model-specific or feature-specific caps at its discretion.

That wording draws the boundary at the center of the case. Anthropic describes the Max multiplier as a session allowance. Kahn argues that customers reasonably understood the plans as delivering five or 20 times Pro's usable capacity over longer billing periods, and that weekly caps made the headline multiples misleading.

The distinction matters most to Claude Code users. Anthropic says activity across Claude's web, desktop and mobile apps and Claude Code draws from the same pool. After subscribers reach the included limits, they can wait for a reset or continue at standard API rates by purchasing usage credits.

Anthropic co-founders Dario Amodei and Daniela Amodei built the company around the promise of reliable AI systems. Its company page describes Anthropic as a public benefit corporation and lists honesty and trust among its operating values. Kahn's case attacks the clarity of a consumer product sitting directly beneath that positioning.

The case remains unresolved

A public docket snapshot shows that Kahn filed an amended complaint on July 20th. On August 7th, Magistrate Judge Alex G. Tse found that another case, Smith v. Anthropic, was related to Kahn's lawsuit. An August 12th order declined to relate two additional cases.

None of those orders addresses whether the Max advertising was false. The court has not endorsed the complaint's calculations, certified a class or found Anthropic liable.

The lawsuit nevertheless reaches beyond a dispute over one $200 subscription. AI companies increasingly sell access through shifting combinations of session windows, weekly ceilings, model-specific budgets and metered overages. A plan name such as "20x" compresses that machinery into a number customers can understand. Kahn's complaint argues that the rest of the machinery can make the number effectively meaningless.