Claude Max usage marketing claims survive dismissal | Magica
Judge lets Claude Max claims over ‘5x or 20x more usage’ proceed
Editorial Team
••📖7 min read
A federal judge found Anthropic’s Claude Max usage advertising plausibly misleading because subscribers could read the multipliers as monthly promises. Contract and misrepresentation claims survive, but several other claims were dismissed with leave to amend.
The court found that monthly subscribers could reasonably interpret the advertised usage multipliers as monthly promises.
Separate weekly caps are central to the dispute; the plaintiffs’ calculations concern Sonnet 4, not every Claude model.
The ruling allows several claims to proceed without deciding liability or class certification.
Anthropic, the company behind the Claude AI assistant, must defend key claims that its Claude Max subscriptions delivered less usage than their advertising promised. In a November 6 order, U.S. Magistrate Judge Alex G. Tse, who is overseeing the proposed class action in the Northern District of California, found the pitch plausibly misleading, while dismissing several other claims with permission to amend.
The dispute concerns what a customer buys when paying a premium for Anthropic’s Claude AI assistant. A promise of “5x or 20x more usage” can describe substantially more capacity within a short session without delivering the same multiple over the subscription’s monthly billing period. Tse found that distinction sufficiently consequential for the case to continue, according to the order.
Anthropic introduced Claude Max in April 2025 after existing subscribers complained about usage limits, according to reporting at launch. The company said expanded access was its most active users’ leading request. Max was a premium subscription to an existing assistant, with promised benefits including higher output limits, early access to new capabilities and priority access during busy periods.
The $200 top tier entered a market where OpenAI already offered a $200 ChatGPT Pro subscription, the launch report noted. More capacity was therefore part of a broader competition for heavy users. It was also costly to provide: the report explained that long Claude conversations and attached documents consumed allowances quickly because the accumulated conversation was processed again with each new turn.
Thailand’s proposed data-centre framework would scale requirements by power demand, screen sites for electricity and water access, and assess domestic benefits. It also covers existing facilities, while conflicting application and approval schedules remain unresolved.
Editorial Team
The plaintiffs were drawn to Claude’s paid plans by Claude Code, a coding tool that can read a codebase, modify files and run tests, according to the order. Karl Kahn uses it for programming tasks; Solomon Smith is a computer-science college student who uses it for coursework, coding projects and a hackathon. They now challenge the advertised comparison with the cheaper Pro plan. The case docket records Kahn’s initial complaint on June 14, 2026, and a consolidated complaint involving both plaintiffs on September 8.
As described in the order, Pro cost $17 to $20 per month, Max 5x cost $100, and Max 20x cost $200. All three paid tiers included Claude Code.
The advertised multipliers applied to five-hour sessions. Separate weekly limits were higher for Max subscribers than for Pro subscribers, but did not carry the same multipliers, according to the complaint as summarized in the order. The plaintiffs’ numerical comparison concerns Sonnet 4, one model in the Claude family.
Challenged plan
Monthly price in the order
Advertised multiplier, applied per five-hour session
Plaintiffs’ calculated weekly usage relative to Pro, for Sonnet 4
Max 5x
$100
5x
3.5x
Max 20x
$200
20x
6x
On those pleaded figures, doubling the subscription price from $100 to $200 bought about 1.7 times the weekly Sonnet 4 allowance: six divided by 3.5. That comparison concerns the plaintiffs’ calculation for one model, rather than a court-tested measurement of every subscriber’s usage or the value of every plan feature.
Kahn alleged that fully using a single five-hour Max 20x session could consume 15% of his weekly allowance. Smith alleged that after upgrading to Max 5x, he had to ration or postpone usage and change how he worked.
Why the help-page disclosure did not end the case
Anthropic argued that its disclosures explained both session and weekly limits. One sales-page asterisk said usage limits applied and that prices and plans could change at the company’s discretion. A linked help page led to another page specifying that the multipliers applied per session.
Tse found the relevant explanation was two hyperlinks away from the primary sales pages, on a page customers did not need to visit to subscribe. Because customers paid monthly, they could plausibly understand the headline usage promise to apply monthly too.
The judge applied existing consumer-advertising precedents involving product labels. An ambiguous but meaningful claim can survive dismissal even when further information appears elsewhere. The linked explanation could still persuade a fact-finder that Anthropic’s advertising was not misleading; it did not dispose of the allegations at this stage.
The same sales process mattered to the contract claim. Anthropic’s terms directed customers to the order process for subscription details. Tse found that those terms plausibly incorporated the usage promises on the signup pages, while the separate help-page qualification plausibly fell outside that process.
Consumer protection has limits here too
Smith’s claim under California’s Consumers Legal Remedies Act survived. That law protects purchases for personal, family or household purposes. Anthropic argued that Claude subscriptions did not qualify as goods or services under it. The court found that Claude Code plausibly qualified as a service because it performed ongoing, customized coding tasks, and that Smith plausibly qualified as a consumer. Educational use did not become commercial use merely because it involved software development.
Kahn’s claim under that law was dismissed because he had not supplied enough detail to establish personal rather than commercial use. Buying an individual subscription did not resolve that question.
The negligent-misrepresentation and breach-of-contract claims also survived. Tse rejected Anthropic’s economic-loss defense to the former, adopting the view that California law permits recovery for negligent misrepresentation as a species of fraud.
But the court dismissed claims under California’s Unfair Competition Law and False Advertising Law. The plaintiffs remained subscribers despite knowing about the disputed marketing, undermining their pleaded basis for an injunction. They also failed to explain why damages would be an inadequate remedy, as required for their restitution request.
Tse separately dismissed the implied-covenant claim challenging changes during billing periods, including tighter capacity limits and moving premium features to cheaper or free tiers. Anthropic’s terms reserved the right to alter features and capacity limits. Subscribers still received more weekly usage than lower tiers and retained Claude Code access. Tse therefore rejected the argument that the changes had deprived the contract of its substance.
The plaintiffs themselves alleged that surging demand strained Anthropic’s capacity. On those allegations, Tse treated tighter limits as an effort to balance supply against demand and found no plausible bad-faith claim. That conclusion concerns the allegations at this stage, rather than an independent measurement of Anthropic’s infrastructure.
That distinction matters: allowing a challenge to a specific advertised comparison does not establish that Anthropic lacked authority to change capacity. Representatives for the company and the plaintiffs did not immediately respond to requests for comment for the contemporaneous report.
Anthropic’s later offers list company-reported allowances of 5x Pro usage per five hours and per week for Max, and 10x weekly usage for Ultra. Ultra’s five-hour limit exemption applies to verified accounts. Source: All AI.
Changed offers, unresolved claims
The challenged plans also need to be distinguished from subsequent offers. A November 7 report on the pricing changes says the tiers became Max and Ultra in late October and that the 20x promise disappeared from the pricing page. According to that report, Anthropic now advertises five times Pro usage per five hours and per week for Max, and ten times Pro’s weekly usage for Ultra, with no five-hour limit for verified Ultra accounts. Those are the company’s newer advertised allowances, rather than independently measured usage.
Those later descriptions do not establish whether the earlier marketing was lawful. Nor does this order decide whether Anthropic is liable or whether the proposed class will be certified.
The plaintiffs have until December 2 to amend their complaint. If they do not, Anthropic must answer the operative complaint by December 16. The unresolved question is whether evidence beyond the pleadings will substantiate the alleged gap between the usage customers were promised and the usage their subscriptions provided.
PJM’s independent market monitor attributes about $11 per megawatt-hour of first-half 2026 wholesale costs to data-center demand in capacity auctions. Energy costs drove more of the overall increase, while construction barriers and an October regulatory rejection complicate the response.
DataVita now says its proposed Lanarkshire AI data centres will be fully grid served, with direct renewable connections only a future possibility. The change in public presentation raises questions about promised power costs and community benefits, although the DV4 planning application already assumed grid supply.
AMD chief Lisa Su calls the chipmaker a data-center-first company as server and AI sales surge. Gaming revenue fell 40%, but PC-chip sales grew, and her remarks leave the supply outlook for expensive Radeon cards unresolved.
A reported U.S. intelligence assessment links Megaspeed executive Alice Huang to China’s Ministry of State Security, raising questions about access to Nvidia’s leadership and chip-export policy. Whether the alleged operation succeeded remains unclear.
Ocient’s $100 million Series C backs tailored AI analytics on an established database platform, with national-security investors and deployment economics shaping its expansion.
Anthropic plans IPO marketing next week and a listing in the week of November 16, according to people familiar with the plans. Buyers are discussing valuations of $1.2 trillion–$1.3 trillion, with some weighing $1.1 trillion, as its filing reveals faster-growing compute costs and extensive infrastructure commitments.
India's IT lobby chief warns that AI could make technology jobs harder for graduates to secure. Hiring is below earlier peaks, but some firms favour AI-equipped beginners over costly experienced staff. The unresolved question is how junior workers will build expertise.
Internet scans and leaked databases expose the spread of ARTEX, a security-testing framework used in cyberattacks. Researchers documented largely autonomous compromises, but the records also show failed cashouts and uncertain counts of independent deployments.
Meta's strategy chief and former CFO is joining Fireworks AI ahead of a reported fundraising push targeting at least a $30 billion valuation. The inference company announced a $17.5 billion valuation in July, making the hire part of a much larger expansion than its older funding figures suggest.
Sand.ai says its $200 million Series A+ round lifts its valuation above $1 billion. Co-lead investor Wangsu plans inference and deployment cooperation, while its filing shows substantial losses behind the startup’s video AI ambitions.
An advocacy group estimates Interior Secretary Doug Burgum’s AI holdings at $18.1 million–$64.4 million as his department advances data centers on public land. His ethics agreement allows retained investments with recusals, while pending permits will determine both the application of those safeguards and the public’s return. Interior denies his decisions are driven by personal investments.
Microsoft is evaluating fewer neocloud partners while expanding its own infrastructure. Delivery-dependent contracts underpin supplier financing, while IREN reports rising revenue alongside a $6 billion funding target for its broader expansion.
Physical Superintelligence says its AI system found a proof of a conjecture limiting quantum query advantages on random inputs. Public proof files enable scrutiny, but do not establish the private system’s cost or repeatability.
OpenAI is expanding its relationship with Hacktron, whose researchers previously reported an attack chain into employee accounts, to test third-party connections with greater access to cybersecurity models.
Google has acquired Crusoe’s southeast Wyoming data center site, where it plans a long-term $17 billion investment. Crusoe describes completed pre-development work; independent reporting points to customer setbacks. Power delivery and carbon capture remain key tests.
The team behind document-parsing service SoMark has released OXR-1.0 code and weights. A benchmark retest supports a strong result, but speed and accuracy were measured separately, and users above funding or revenue thresholds need a commercial license.
OpenAI is reportedly developing a Business Hub with daily mention and click charts, product updates and business agents. The unannounced feature could give verified businesses direct visibility into ChatGPT responses, without revealing prompts or competitor rankings.
Palantir and Nvidia have specified an integrated stack for running AI in customer-controlled environments. The architecture builds on existing software and infrastructure offerings, with published configurations spanning 32 to 640 GPUs, while pricing and workload economics remain unproven in the launch materials.
Democrats’ incoming congressional majorities put hearings and investigations within reach, but AI rules and crypto legislation still face bipartisan bargaining and a Republican president.
