Anthropic Sued Over Claude Max Plans That Overpromise Usage
A class action filed June 14, 2026 claims Anthropic's $200/month Claude Max 20x plan delivers only 6-8x Pro usage, not the advertised 20x. Here's what we know.
A proposed class action filed on June 14, 2026 in the U.S. District Court for the Northern District of California alleges that Anthropic's Claude Max plans deliver far fewer usage credits than their names suggest. Plaintiff Karl Kahn paid $200 a month for the Max 20x tier to run Claude Code through extended coding sessions, then found the plan behaved more like a 6-to-8-times plan. Anthropic has declined to comment, and no court has ruled on the claims. The case is active, with an amended complaint filed July 20, 2026.
What happened
| Detail | Fact |
|---|---|
| Case name and number | Kahn v. Anthropic, PBC, 3:26-cv-05763 |
| Filed | June 14, 2026, Northern District of California |
| Amended complaint | July 20, 2026 |
| Latest court activity | Related-case order from Magistrate Judge Alex G. Tse, August 12, 2026 |
| Plan at issue | Claude Max 20x at $200/month |
| Alleged actual multiplier | 6-8x Pro, per Engadget reporting on the complaint |
| Also named | Claude Max 5x at $100/month, alleged to deliver ~3.5x |
Kahn bought the top-tier Claude Max plan specifically to run long Claude Code sessions without hitting usage limits. He alleges that one five-hour Claude Code session consumed roughly 15% of his weekly allowance, which is math that does not square with a “20x” label on a $200-a-month product.
The core of the complaint rests on guidance Anthropic sent subscribers in July 2025. According to figures reported by both TechCrunch and Engadget, Anthropic told users to expect the following weekly Claude Code hours:
| Plan | Price/month | Sonnet 4 hours/week | Opus 4 hours/week |
|---|---|---|---|
| Pro | $20 | 40-80 | Not included |
| Max 5x | $100 | 140-280 | 15-35 |
| Max 20x | $200 | 240-480 | 24-40 |
Compare the Sonnet 4 ceilings: 80 hours for Pro, 480 for Max 20x. That ratio is 6x. Kahn’s argument is straightforward: the plan is called “20x” but the published numbers do not support a 20x interpretation by any simple reading.
The double-limit problem
Anthropic’s support pages describe two separate constraints. First, a five-hour session limit. Second, a weekly usage ceiling across all models. TechCrunch reported that Anthropic announced those weekly caps on July 28, 2025, with enforcement starting in late August, after the company said some subscribers were running Claude Code continuously in the background. Anthropic said at the time it expected the caps to affect fewer than 5% of subscribers.
Two independent limits are not inherently unfair, but they create a trap. A user can exhaust a session limit, wait for the reset, and then discover the weekly cap is still active. The multiplier on the plan name does not tell you which limit bites first or how the two interact.
Why it matters
This lawsuit is not just about one subscriber’s frustration. It asks a question every AI company selling flat-rate tiers has quietly avoided: does a plan name like “20x” constitute a specific, enforceable promise about usage, or is it just a rough marketing label?
Anthropic is not alone in facing this tension. According to TechCrunch, Anysphere changed Cursor’s Pro pricing in June 2025 and later apologised for how poorly it communicated the change. Replit made similar adjustments. The pattern is consistent: developers will use as much compute as the product allows, pricing teams get surprised, and communication suffers.
If a federal court allows Kahn’s claims to proceed, every AI company selling a named usage multiplier will need to audit whether that name can survive a customer’s spreadsheet. Businesses using AI integration tools that rely on Anthropic’s API through subscription tiers should pay attention to how usage limits are described in their contracts and vendor terms, not just the plan names.
For context, Anthropic has also been named in a separate copyright suit filed August 28, 2026, by Sony Music Publishing and Warner Chappell. That case, covered in our earlier piece on Sony and Warner’s lawsuit over Claude training data, is unrelated to the Max plan complaint but adds to the legal pressure on the company.
Our take
No court has found Anthropic guilty of anything. The lawsuit is an allegation, and Anthropic may well have a strong defence. But the underlying problem is real and predates this case.
Naming a plan “20x” is a choice. If the actual Sonnet hour ratio tops out at 6x by your own published guidance, that gap will eventually produce exactly this kind of dispute. Usage caps and rate limits are legitimate tools for managing infrastructure costs. The mistake is not having them. The mistake is dressing them in a number that most customers will read as a ceiling guarantee rather than a usage guideline.
Businesses that pay premium prices for AI tools, whether for coding, content, or workflow automation, should read the fine print on session limits and weekly caps before committing to annual contracts. Ask the vendor to define, in writing, what the multiplier means in terms of hours, tokens, or requests per week. A clean answer tells you a lot about how that vendor thinks about its customers.
What to do about it
- Pull your Claude Max usage logs now and check whether your weekly hours match the ranges Anthropic published (40-80 hours Sonnet for Pro, 240-480 for Max 20x).
- Ask Anthropic support to confirm, in writing, which limit resets first and how the session cap and weekly cap interact for your tier.
- If you use Claude Code for production work, test your actual throughput over a full week before renewing a Max plan at the next billing cycle.
- Compare your real usage data against the advertised multiplier. If the gap is substantial, you have grounds to request a plan review or downgrade before your next payment.
- Watch the docket for Kahn v. Anthropic, PBC (3:26-cv-05763). A decision on class certification would be the next significant milestone.
The case is still in early stages, but the usage math is already public. That alone is worth checking against your own bill.
Frequently asked questions
What is the Kahn v. Anthropic lawsuit about?
Karl Kahn filed a proposed class action (case 3:26-cv-05763) on June 14, 2026, claiming Anthropic's Claude Max 20x plan delivers only around 6-8 times the usage of the Pro plan, not the 20 times suggested by its name. He paid $200 a month for the plan to run Claude Code without hitting limits.
How much does Claude Max cost and what does it include?
Anthropic lists Claude Max 5x at $100 a month and Claude Max 20x at $200 a month. The 20x plan is described as providing 20 times more usage per session than the $20 Pro plan, with weekly Sonnet 4 hours of 240-480 and Opus 4 hours of 24-40 according to Anthropic's July 2025 guidance.
Has Anthropic been found guilty of misleading customers?
No. The lawsuit is an allegation, not a verdict. No court has ruled that Anthropic misled anyone. The case was still active as of August 12, 2026, with no settlement or merits ruling on the public docket.
What are Claude Max's actual usage limits?
Anthropic's support pages describe two limits: a five-hour session limit and a weekly usage ceiling across models. Weekly Sonnet 4 hours are 240-480 for Max 20x versus 40-80 for Pro, a ratio of 6x at the ceiling, not 20x.
