Subscriber Sues Anthropic, Alleging Claude Max Plans Deliver a Fraction of Advertised Usage

What Was Promised vs. What Happened

Anthropic's Max subscription tiers were built on a simple pitch: pay more, get more. The Max 5x plan, at $100 per month, advertises five times the usage of the base Pro package. The Max 20x plan, at $200 per month, promises 20 times the usage. Claude Pro itself costs between $17 and $20 per month.

According to CNET, the lawsuit filed in the US District Court for the Northern District of California alleges the reality is considerably less generous. The complaint claims the Max 20x plan actually delivers "just six to eight times the usage of Pro," and the Max 5x plan delivers "just three-and-a-half times the usage of Pro." If those figures hold up, subscribers paying $200 a month are getting less than half what Anthropic advertised.

Who Filed the Suit

The plaintiff is Karl Khan, a Washington DC-based resident. Per Engadget, Khan started using Claude for personal reasons, then upgraded his subscription twice as he shifted to coding work, landing on the Max 20x plan by April 2026. He said a single five-hour coding session consumed roughly 15 to 20 percent of his entire weekly allocation.

Khan said he compared his experience across subscription tiers and concluded the gap between what Anthropic advertises and what it delivers is systematic. The lawsuit is seeking class-action status on behalf of all US consumers who purchased a Max plan since Anthropic launched the tiers in April 2025.

Why This Complaint Has Technical Roots

Anthropic's own website acknowledges usage is variable. The company explains that message limits depend on "message length, including the length of files you attach, the length of your current conversation, and the model or feature you use." That language gives Anthropic wiggle room: usage is not a fixed number of messages, it's a compute budget.

Large language models run on tokens. Every prompt converts words, punctuation, and character groups into numerical inputs the model processes. Complex coding prompts, especially those using Claude Code, burn through tokens at a far higher rate than a simple question. Engadget noted that Anthropic imposed weekly rate limits on Claude Code in July 2025 specifically because some users had been running the coding agent around the clock.

This structural tension sits underneath the lawsuit. Traditional software subscriptions give you a fixed seat or a fixed storage cap. AI inference costs scale with what you ask the model to do. A user running intensive coding sprints will hit a wall that a casual chat user might never see.

The Strongest Counter-Argument

Anthropic's most defensible position is that no AI company has yet cracked the problem of translating compute costs into consumer-friendly usage promises, and that the variability language on its website does put users on notice. If Claude Code consumes tokens at a dramatically higher rate than standard chat, a heavy coding user comparing their experience to a lighter user's may be drawing an unfair apples-to-oranges comparison. Anthropic could also argue that "five times the usage" refers to a baseline session type, not worst-case intensive coding workflows.

That argument runs into a problem: Anthropic actively markets Max plans to coders and specifically positions Claude Code as a key use case. Selling a product to heavy users while burying the caveat that heavy use is exactly what drains the budget fastest is a harder position to defend in court than in a marketing deck.

What the Lawsuit Actually Claims

The complaint, as reported by both Engadget and CNET, does not allege fraud in the criminal sense. It alleges false advertising and consumer deception under civil standards. The specific numbers Khan's lawyers put forward—that the 20x plan delivers 6-8x and the 5x plan delivers 3.5x—are derived from Khan's own comparative testing across tiers. Those are allegations, not adjudicated facts. No investigation or charges have been filed by any regulatory agency.

Anthropic declined to comment to Engadget. The company has not released a public statement addressing the specific numbers in the complaint as of June 15, 2026.

The Broader Pattern

Benzinga noted this suit lands as Anthropic is navigating the broader tension between venture-backed growth and the actual economics of running large models. Engadget pointed out that rate limit complaints have been a recurring thread on Reddit, with users frequently reporting that a single Claude Code session wipes out a substantial portion of their weekly budget.

The disconnect is not unique to Anthropic. The entire consumer AI subscription market is built on pricing models that were not designed with LLM inference costs in mind. But Anthropic is the one with a lawsuit now.

What Comes Next

The case was filed in the Northern District of California. For it to proceed as a class action, Khan's legal team will need to demonstrate that the alleged shortfall in usage is consistent across subscribers, not just a product of his own particularly intensive coding sessions. That will likely require Anthropic to produce internal data on token consumption across its Max tier user base, which is exactly the kind of discovery Anthropic will want to avoid.