Anthropic sued over alleged Claude Max subscription limits

Karl Kahn, a Claude subscriber in Washington, D.C., has filed a proposed class action asking a federal court to make Anthropic reimburse buyers of its two Max plans. He says the $100 Max 5x and $200 Max 20x tiers deliver much less than the usage multipliers they are sold on. The Wall Street Journal reported the suit on August 31.
The reports give different filing days. The Journal says the suit was filed on a Monday, and a second report says Sunday. Kahn’s lawyers at Vaca Daffan Law wrote in a June 15 post that they filed the complaint in the US District Court for the Northern District of California “last night.” Vaca Daffan is working with California co-counsel Elliot Conn and Colin Hector, an FTC alum, both of Conn Law, PC. The case is captioned Kahn v. Anthropic, PBC.
What the complaint counts
The complaint rests on two ratios. Max 20x is sold as 20 times the usage of Pro, but the complaint says it actually comes to “just six to eight times the usage of Pro.” Max 5x, it says, delivers “just three-and-a-half times the usage of Pro” instead of five. Pro costs between $17 and $20 a month, depending on billing.
The filing also lays out how Kahn got there. He started using Claude for personal tasks, moved on to coding work and upgraded twice, reaching Max 20x by April 2026. He says a single 5-hour coding sprint regularly used up nearly 20% of his weekly allowance. After comparing his experience across the tiers, he concluded that what Anthropic advertised and what it delivered did not match.
The broader charge is about how usage is counted. The complaint calls Anthropic’s website “a black box, without any meaningful description of how usage is calculated,” and alleges that the company never clearly defines what counts as a single session. Subscribers who hit their limit on a premium plan are prompted to buy credits to keep working. The filing also cites customer complaints, saying “many subscribers have reported their frustration and complained that [Anthropic] has engaged in a bait and switch.” Similar stories fill at least one popular Reddit thread, whose original poster called the premium plan “extremely misleading.”
Which limit does 20x describe?
Every Claude plan runs on two meters: a limit per 5-hour session and a weekly cap. Anthropic’s blog posts explain that both apply and give advice on managing usage for big projects. None of them says how many prompts a given plan allows in either window. On the pricing page, the offer reads “Choose 5x or 20x more usage than Pro*.” The footnote says only “Usage limits apply,” and another line says prices and plans “are subject to change at Anthropic’s discretion.” In a blog post, Anthropic also reserves the right to “limit your usage in other ways, such as weekly and monthly caps or model and feature usage, at [its] discretion.”
An X user, SataEricUX, says the 20x applies only to the 5-hour window and not to the weekly limit. A Hacker News poster who passed that claim on said they had assumed the $200 plan bought four times the usage of the $100 one, and called the labelling “very misleading.”
The limits have also moved during the period the suit covers. On May 13, 2026, Anthropic started a promotion that raised Claude Code weekly limits on Pro, Max and Team by 50% and left the 5-hour limits unchanged. That temporary increase has since been extended until September 13.
Who would be in the class
The suit asks for class status on behalf of anyone who bought and used Max 5x or Max 20x since April 2025. Anthropic announced Max on April 9, 2025, as a plan for daily users who kept hitting Pro’s limits, so the proposed class reaches back to launch. The restitution sought has not been put at a figure, but the complaint says the “amount in controversy exceeds the sum of $5 million exclusive of costs.”
On September 14, Anthropic will permanently raise standard weekly limits by 25% for Pro, Max, Team and seat-based Enterprise plans. Testing Catalog points out that the increase is measured against the standard limit and not the promotional one, which leaves subscribers about 17% below what they have during the promotion.
Up next
Claude cheated on 2.4% of its own safety runs
Anthropic’s monitor caught Claude “exfiltrating test labels from a remote API and cherry-picking results” in 2.4% of about 1,600 research transcripts, even as its fixes improved all 10 alignment benchmarks.
Related stories
- Weekly cap on Claude Max lands in an expanded suit
- Weekly limits rise 25% permanently on September 14
- The same Claude Code work can cost 40x more on the API than on a subscription
- Claude usage credits bought from Dec 7 expire after a year
- Claude Code subagent status updates cost about 99% less
- Claude Ultra drops the 5-hour cap but keeps the weekly one
Comments
No comments yet. Be the first.
Join the conversation
Sign in to leave a comment. It appears after moderation.

