Wait, You Paid $200 For Claude Max And Got Rationed Like It’s A Seder? Anthropic Is Being Sued

by

Oh, fantastic. Anthropic is now facing a proposed class-action lawsuit because apparently, if you shell out up to $200 a month for their fancy Claude Max plans, you don’t get anywhere near the usage they were out here promising like it was a done deal. What is this, a bagel shop that says you get a dozen lox bagels for $20 and then hands you three and a half and says “the rest are for next week, buddy”? Come on.

these premium tiers cost $100 and $200 a month, and Anthropic has been marketing them as giving you five times and 20 times the usage you get from their standard Pro plan. Five times! Twenty times! That’s what they’re saying! What are they doing, counting usage in “Claude dollars” that only exist on their server? That’s not how multipliers work!

But according to the lawsuit, the actual limits they slap on subscribers are way, way lower than those big shiny multipliers they’re touting, and they’re nearly impossible to predict until you’re already staring at a usage cap message while you’re in the middle of a coding project. It’s like ordering a large popcorn at the movies and getting a medium, but you don’t find out until you’re already 10 minutes into the movie and you’ve eaten half the bag. Rude.

Kahn says he upgraded to the Max 20x plan specifically because he was using Claude more and more for software development and coding work, which makes sense, that’s what these things are for, right? Well, according to the filing, a single five-hour work session ate up roughly 15% of his entire weekly usage allowance. Fifteen percent! For five hours! That’s like paying for an all-you-can-eat buffet, eating one plate of salad, and being told you’ve used up 15% of your food quota. What is this, the Soviet Union?

Look, The Whole Problem Is These Stupid Usage Caps

To back up their claims, the lawsuit points to emails Anthropic supposedly sent to subscribers back in July 2025 that spelled out exactly what weekly usage allowances you got for each Claude model and subscription tier. Kahn’s argument is that these emails are basically proof there’s a huge gap between how they were marketing these plans and the access people were actually getting. It’s not even a he-said-she-said situation! They wrote it down! Who does that?

The filing is asking the court to rule that Anthropic’s marketing practices were either misleading or flat-out fraudulent, and they’re asking for some kind of relief for everyone who got scammed into paying for a plan that didn’t deliver. Of course, this all drops right as Anthropic is gearing up for a highly anticipated public offering, because nothing says “we’re a solid, trustworthy company” like getting sued for lying to your customers the week before you go public. Perfect timing, guys. Really.

Oh, And This Isn’t Even The Only Mess They’re Dealing With

This lawsuit is just the latest headache for Anthropic, by the way, coming just days after they got dragged into a whole other controversy over access to their advanced AI models. It’s like when you have a bad cold, and then you spill coffee on your favorite shirt, and then you realize you left the stove on. Just one thing after another, am I right?

As crypto.news previously reported, Anthropic cut off access to their Fable 5 and Mythos 5 models after they had to comply with a U.S. government directive tied to export controls. You know, the kind of rules that make you lock down your fancy toys so people from other countries can’t use them. Very dramatic.

Anthropic said the order required them to put restrictions on foreign nationals, including foreign-national employees who work for them both inside and outside the United States, so they just shut off access to both those models entirely while keeping all the other Claude models up and running. I guess when the government says jump, you just shut down half your product and say “how high?” No questions asked.

CoinFund founder Jake Brukhman piped up recently to say that advanced AI models have basically become the ultimate point of centralized control these days. His whole thing is that decentralized AI networks are starting to look a lot more appealing, because instead of all the big computing power being hoarded by two or three companies that can turn off your access whenever they feel like it, you can spread that compute out across a bunch of different people. It’s like instead of one guy owning the entire bagel bakery and deciding who gets a bagel, everyone in the neighborhood has a little part of the bakery and you can just get your bagel whenever you want. Makes sense, right?

He named a bunch of projects working on this stuff-Gensyn, Prime Intellect, Bagel, Pluralis, Nous Research, Macrocosmos AI, Covenant-as examples of teams building these distributed AI training systems. He says it looks like these decentralized efforts can actually go toe-to-toe with the centralized big players, though he admits there are still a bunch of technical kinks to work out. Hey, Rome wasn’t built in a day, and neither was a bagel that doesn’t get rationed by a random AI company.

Anyway, this new lawsuit is putting Anthropic’s whole subscription business under a whole different kind of microscope right now, as they’re already trying to juggle both regulatory headaches and angry customers who feel like they got scammed. Tough break, really. Maybe next time don’t promise 20 times the usage if you’re just gonna give people 3 times and tell them to “ration their prompts” or whatever.

Read More

2026-06-15 21:38