Anthropic faces class-action claim over Claude Max 20x limits

Why it matters

The case targets the gap between simple AI subscription marketing and the variable compute economics behind heavy Claude usage.

Anthropic, the AI lab co-founded by Dario and Daniela Amodei, is facing a proposed class-action lawsuit over whether its $100 and $200 Claude Max subscriptions delivered the usage levels Anthropic sold to heavy users.

The complaint, filed June 14 in the U.S. District Court for the Northern District of California by Claude subscriber Karl Kahn, accuses Anthropic of misleading customers by marketing Claude Max 5x and Claude Max 20x as offering five and 20 times the usage of the lower-priced Claude Pro plan. Engadget reported the lawsuit on June 15.

Kahn's core allegation is arithmetic, not model quality. The lawsuit says Claude Max 20x delivers "just six to eight times" the usage of Pro, while Max 5x delivers "just three-and-a-half times" Pro. It also challenges Anthropic's claim that the $200 Max 20x tier offers a 50% savings over the $100 Max 5x tier.

Anthropic's own public product copy gives the case its hook. On Anthropic's current Claude pricing page, the Max plan is described as offering "5-20x more usage than Pro" and gives subscribers the option to choose "5x or 20x more usage than Pro." Anthropic's Max plan help page says Max 5x provides five times more usage per session than Pro and Max 20x provides 20 times more usage per session than Pro, while also noting that Max plans have weekly usage limits and that Anthropic may impose other caps by model, feature or time period.

That dual framing is the legal and commercial tension. Anthropic sells simple multiples to consumers, but it meters Claude through a variable system of session limits, weekly limits, model-specific usage and token-heavy features. Anthropic's usage limits documentation says usage depends on conversation length and complexity, selected model, feature choice and effort level, and that usage across Claude's web app, Claude Desktop and Claude Code counts toward the same limit.

Kahn says he bought into the higher tiers for coding work. According to the complaint, he subscribed to Claude Pro around June 23, 2025, upgraded to Max 5x on January 5, 2026, and moved to Max 20x on April 21, 2026 after using Claude Code extensively. Within weeks, he says, he was hitting weekly limits. The complaint says one full five-hour window could consume 15% of the weekly allotment Anthropic assigned to his Max 20x plan.

The proposed class covers U.S. customers who bought or upgraded to Claude Max 5x or Max 20x through Claude.com or the Claude desktop app from April 9, 2025, the day Anthropic introduced Max, to the present. The complaint seeks damages, restitution, injunctive relief and other remedies under California consumer-protection and false-advertising law, along with claims for negligent misrepresentation, breach of contract and unfair competition.

Anthropic introduced Claude Max on April 9, 2025, pitching it as a plan for users who needed expanded access for demanding projects. The launch post said the top request from Anthropic's most active users was more Claude access and set two tiers: $100 a month for 5x more usage than Pro and $200 a month for 20x more usage than Pro.

The timing matters. Anthropic was turning Claude from a chat product into a developer work surface, with Claude Code bringing the model into terminals and software projects. That made Claude more valuable to programmers, but also more expensive for Anthropic to serve. A short consumer chat and a long coding session against a large codebase can look similar in a subscription checkout flow and very different in GPU cost.

Anthropic's current workaround is to push users who hit limits into paid overage. Its usage credits page says Pro, Max 5x and Max 20x subscribers can continue after reaching included limits by switching to standard API-rate billing, with those charges separate from the subscription. That may be rational pricing for compute-intensive work. It is also exactly why a customer who paid $200 for "20x" may feel the product was sold in consumer language and delivered in infrastructure language.

The lawsuit does not prove Anthropic underdelivered. It frames a narrower question: when an AI company advertises a usage multiple, does that promise mean a simple relationship between plans, or can the company satisfy it through a more complex, shifting system of token budgets, model limits and time windows?

For Anthropic, the risk is bigger than refunds on one subscription tier. The Amodeis built Anthropic around trust, safety and reliability, and Claude's premium plans are part of the company's push to turn expensive frontier models into recurring consumer and developer revenue. If a court forces more precise disclosure of usage math, it could make AI subscription pricing less like SaaS and more like cloud computing: metered, conditional and harder to market in a single number.