A subscriber of Claude has taken legal action against Anthropic, claiming that the company has misrepresented the usage restrictions associated with its Max plans. According to recent reports, Karl Kahn, a resident of Washington DC, filed a federal lawsuit alleging that Anthropic exaggerated the amount of access users would have to Claude. The suit contends that the actual usage limits for the Max 5x and Max 20x plans are difficult to ascertain and are notably lower than what the company advertises on its website and other platforms.
Anthropic provides three paid subscription options for users. The entry-level package, known as Claude Pro, is priced at $17 per month and claims to offer a usage level that is five times greater than what the free service provides during peak hours. The company states that the allowable number of messages may fluctuate based on various factors, such as the length of the messages, any attached files, the ongoing conversation length, and the specific model or feature utilized. Since April 2025, the company has also rolled out the Max 5x and Max 20x plans, which promise up to five and twenty times the usage limits compared to the Pro plan at costs of $100 and $200 monthly, respectively.
The lawsuit asserts that the actual limits provided by the Max 5x and Max 20x plans fall significantly short of the advertised figures. Following his initial use of Claude Code, Kahn upgraded to the Max 20x plan in April, only to quickly reach his weekly limits. In one instance, during a five-hour session, he used up 15 percent of his weekly allocation. The lawsuit aims for class-action status for other U.S. customers who have subscribed to the Max plans since they were introduced last year.
Anthropic has opted not to provide comments regarding the lawsuit.
The limitations of Anthropic’s services have sparked discussions among users on Reddit, with some expressing frustration over exceeding their assigned limits with minimal usage. Last July, the company implemented weekly limits on Claude Code to address complaints from users running the coding agent continuously in the background.
Large language models function using a system known as tokens. When a user inputs a query into Claude’s prompt, it translates words, characters, and punctuation into numerical values that correspond to distinct patterns learned during the model’s training. Generating a response incurs both input and output costs, which can vary significantly based on the complexity of the inquiry.
This lawsuit underscores a growing discrepancy between what consumers expect and the true expenses associated with AI processing. Conventional software subscription models do not align well with the computational demands of large language models. Currently, venture capital funding is helping to absorb these costs, a situation that may become even more pressing as Anthropic and OpenAI prepare for public offerings.
