What happened
In an order on Anthropic’s motion to dismiss, the U.S. District Court for the Northern District of California considered claims by Karl Kahn and Solomon Smith, who allege that Anthropic misleadingly marketed Claude Max subscriptions. The plaintiffs say the company’s “5x or 20x more usage than Pro” representations could lead customers to expect that amount of additional usage over a monthly billing period. Anthropic argued that the claims should be dismissed.
The court concluded that the advertising could plausibly mislead a reasonable consumer at this stage of the case. It reasoned that customers seeing monthly subscription prices alongside the multipliers could reasonably infer a monthly comparison. The order describes the relevant plans as Pro at $17 to $20 per month, Max 5x at $100 per month, and Max 20x at $200 per month. The court also noted that the paid plans include access to Claude Code, which it described as an agentic coding system.
Anthropic’s usage disclosures, as described in the order, say the multipliers apply per session: Max 5x provides five times the usage per session of Pro, and Max 20x provides 20 times the usage per session. Those disclosures also state that both per-session and weekly limits apply. The court found that the linked explanation was two hyperlinks away from a primary sales page, and that the page’s “Usage limits apply” disclaimer did not make it impossible for customers to plausibly read the multiplier as applying monthly.
The order discusses weekly limits separately. It cites a complaint calculation for Claude Sonnet 4 under which Max 5x users receive 3.5 times Pro’s usage per week and Max 20x users receive six times Pro’s weekly usage. A user can therefore have greater capacity in a five-hour session while still reaching a weekly limit sooner than expected. These figures are allegations cited by the court, not a new usage audit conducted by the court.
The ruling is mixed. The court denied dismissal of Smith’s claim under California’s Consumer Legal Remedies Act (CLRA), finding that he plausibly alleged personal use as a student. It dismissed Kahn’s CLRA claim with leave to amend because he had not sufficiently described personal rather than commercial use. The court dismissed the plaintiffs’ California Unfair Competition Law and False Advertising Law claims, also with leave to amend, on grounds involving standing and available remedies. The order further addresses negligent misrepresentation and contract claims; this ruling decides whether pleaded claims can proceed, not whether Anthropic is ultimately liable.
What the archive adds
New Runtime’s earlier coverage reported that Anthropic renamed its former Max 20x subscription as Ultra while keeping its $200 monthly web price (October 31, 2026 note). This court order is a new event in the same subscription story: it concerns the legal dispute over what the usage multipliers communicate, rather than the plan’s name or price.
Trend signal
This is adjacent to New Runtime’s agent-ready software exposes capabilities pattern, but it is not a direct demonstration of capability design. The dispute instead highlights that access to an agentic coding service is sold through bounded plans whose limits shape practical use. The case may sharpen scrutiny of how those boundaries are communicated, but it does not establish a broader product-design trend on its own.
Why now
The ruling advances the case after Anthropic sought dismissal. The court’s focus on the distance between a headline usage claim and the detailed session-limit disclosure makes the subscription presentation itself a live issue in the litigation, rather than a settled question about plan naming or price.
Why it matters
For teams buying AI services, a multiplier is not a complete measure of usable capacity unless its comparison period and limit structure are clear. A plan can support a much larger burst of work in one session while offering a smaller increase over a week, which matters for sustained coding workflows and budgeting. The order does not establish that Anthropic’s advertising is unlawful; it says the plaintiffs’ interpretation is plausible enough that the relevant claim should not be dismissed now.
For companies selling agent access, the practical lesson is about the product boundary as well as legal exposure. Pricing pages, usage explanations, and limit documentation need to describe the same unit of comparison in a way customers can see before purchase. Builders choosing a plan should distinguish per-session headroom from weekly capacity and check how each limit applies to their workload.
Limits and uncertainty
This is a ruling on a motion to dismiss, not a trial verdict or a finding that consumers were actually deceived. The court accepted the complaint’s factual allegations for purposes of the motion and evaluated whether the claims were plausible. The weekly Sonnet 4 ratios cited in the order come from the complaint’s calculation; the ruling does not independently measure usage across subscribers or models. Several claims were dismissed with leave to amend, and the case may develop through amended pleadings and later proceedings. The order does not settle how Anthropic’s disclosures will be assessed on a fuller factual record.
Reader takeaway
A “5x” or “20x” allowance can describe a session rather than a month, and the weekly ceiling can tell a different story again. When comparing AI subscriptions for sustained work, inspect both the time window and the relevant cap; headline multipliers alone do not reveal how much work a team can complete over a billing cycle.