Claude's Max 20x is now Ultra at the same $200 — and the weekly multiple is 10x, not 20x

Anthropic’s consumer plan ladder changed some time in the last twenty-four hours, and the only place it is documented is a help-centre article with no date on it.

On 29 October, claude.com/pricing listed Free, Pro, Max 5x at $100 and Max 20x at $200. Today it lists Free, Pro, Max at $100 and a new plan called Ultra at $200. The help centre explains the mechanism in a section titled “What changed from Max 20x?”:

“The Max 20x plan is now called the Ultra plan. Existing Max 20x subscribers move to Ultra automatically, with no price change.”

A rename with no price change. Two sentences, and both are true as far as they go. What neither says is that the number in the old plan’s name was twenty, and the number the new plan publishes is ten.

The number in the name

Ultra’s usage figure appears three times in Anthropic’s own article, phrased identically each time: “Get 10 times the Pro plan’s usage every week.” The pricing page agrees — its comparison table gives the weekly row as Free “Limited”, Pro “1x”, Max “5x”, Ultra “10x”.

The old plan was called Max 20x. Anthropic no longer publishes what Max 20x’s multiple was on any reachable page; the word “20x” survives in exactly two places, both vestigial. One is the heading of the article explaining the rename. The other is the Ultra card’s signup link on the pricing page, which still points at claude.ai/upgrade/max?plan=max_20x — the old plan identifier, carried forward under a new label.

That is the evidential position, and it should be stated honestly rather than dressed up: the “before” figure rests on the plan’s own name, on Anthropic’s statement that this plan is that plan renamed, and on this publication’s own verification of the pricing page on 29 October. Anthropic has not disputed the twenty; it has simply stopped publishing it.

What the buyer gets back

The trade is not nothing. Ultra’s other headline benefit is the removal of Claude’s five-hour rolling window:

“No five-hour limit for verified accounts: Work at your own pace within your weekly limit.”

Max keeps that window — “five times the Pro plan’s usage every five hours and every week” — and so does Pro. Ultra drops it. For anyone whose working pattern was to hit the five-hour wall mid-task and wait, that is a genuine upgrade, and arguably the more valuable half of the package. A weekly ceiling you can spend however you like is worth more than a higher weekly ceiling rationed into five-hour slices.

So the honest summary of the change is not “a cut”. It is a reshaping: the session cap disappears and the weekly cap halves, at an unchanged $200. Bursty users win. Sustained heavy users lose. Anthropic has published the first half and not the second.

The condition on the headline benefit

The removal of the five-hour limit carries a qualifier that is easy to miss, because the pricing page renders it as a dagger footnote: “No 5-hour limit for verified accounts.”

Verification is not something a buyer does. Anthropic’s wording:

“we automatically verify Ultra accounts using factors like your account and payment history”

and, for everyone else, “If your account isn’t verified yet, a temporary limit applies.” There is no application, no published criteria beyond “account and payment history”, and no stated timeline — only “We remove it automatically once your account is verified, so there’s nothing you need to do.” The article pre-empts the obvious workaround too: “Verifying your phone number at checkout is a separate step and doesn’t remove this limit.”

The practical shape of this: a new customer can pay $200, receive the lower weekly multiple immediately, and not receive the benefit it was traded for until Anthropic’s systems decide they qualify. The self-check is Settings → Usage — if there is no “Current session” bar, the account is verified. Nothing on the purchase page tells a prospective buyer which side of that line they will land on.

The credits are real money and they are not capacity

Ultra’s third listed benefit is “$200 a month in API credits, up from $100 on Max”. At face value that is a 1.00x return on the subscription price, which is the best ratio in consumer AI — and Anthropic now publishes the Team split that had dropped off the pricing page: $20 per Standard seat and $100 per Premium seat, pooled into one balance and capped at $500 a month, with a worked example showing three Standard and two Premium seats receiving $260.

The restriction is the one that matters, and Anthropic states it plainly:

“The credits don’t change your usage limits in Claude, Claude Code, or Claude Cowork.”

They also do not cover interactive Claude Code in the terminal, IDE, desktop or web, nor extra usage once a plan limit is hit. They require seven days on the plan, expire at the end of each billing cycle with no rollover, are spent before purchased credits, and can be linked to exactly one Console organisation — which the subscriber cannot change without contacting support. Claude Code buyers in particular should read that first sentence twice: the credits are a parallel API budget, not agent headroom, and they are worth $0 to anyone without an API workload.

No announcement, anywhere

What makes this hard to price is not the change. It is that the change is undocumented.

There is no post about the Ultra plan on anthropic.com/news, which currently carries nothing of the kind back through mid-September. There is no entry in Anthropic’s own Claude release notes, whose most recent items are Claude Duet on 21 October and Haiku 5.5 on 7 October. The release notes still carry an item headed “Monthly API credits for Max and Team plans” — pre-Ultra naming — linking to an article that is now titled “Monthly API credits for Max, Ultra, and Team plans”. Both support articles show only the relative string “Updated today”; neither carries an absolute date, a launch date or an effective date. The help-centre collection has been retitled “Pro, Max, and Ultra plans” and contains one Ultra article.

A plan tier that costs $2,400 a year changed what it includes, and the complete documentary record is one undated help-centre paragraph.

The same $200, the other direction, the same silence

Twenty-four hours earlier, the other $200 tier in consumer AI moved the opposite way.

OpenAI had signalled a reduction in the included allowance for ChatGPT Pro 200 — a claim this publication twice failed to verify against OpenAI’s own pages, on 21 and 30 October, and twice marked as reported-not-confirmed for that reason. On 29 October, a named OpenAI staffer, Thibault Sottiaux, posted that the company is “doubling all usage limits for all Pro 100/200/500 tiers and Business Premium, effective today,” and that it “won’t be reducing usage limits for Pro 200, so your Work and Codex usage will stay the same.”

That reversal is, as of 30 October, on no OpenAI-owned page either. learn.chatgpt.com/docs/pricing prices Pro as “From $100” with “Plans at $100, $200, or $500 USD per month”, publishes no allowance multiple for any plan, states only that “Pro plans currently have no five-hour limit”, and carries no notice dated 29 or 30 October. OpenAI’s What’s New page, whose newest section is headed “October 26–30, 2026”, lists an iMessage beta and GPT-6.1 Terra — no usage-limit item. The two help-centre articles that would carry it return 403.

So inside one 48-hour window: OpenAI doubled the allowance on its $200 tier according to an employee’s social post and nowhere else, and Anthropic halved the published weekly multiple on its $200 tier according to an undated help article and nowhere else. At the top of the consumer market, the allowance behind the price is now the least documented number in the product. The price is published, stable and easy to compare. What it buys is published inconsistently, changed without notice, and in OpenAI’s case not published at all.

What to do about it

For anyone choosing between $100 and $200 of Claude, the question has genuinely changed shape. It is no longer “how much more usage do I get” — it is which limit binds you. Max retains the five-hour window and 5x weekly. Ultra removes the window and offers 10x weekly. Bursty, session-shaped work favours Ultra more than it used to; steady high-volume work favours it considerably less than the old 20x name implied. Our Claude review and the best AI chatbots guide carry the revised ladder, and the pricing tracker records the change with a dated event.

Two structural points make the decision lower-stakes than it looks. Neither Max nor Ultra is billable annually — both are monthly-only — so there is no discount to forgo and no lock-in to fear, and switching after a month of real use costs nothing but the difference. And the API credits, generous as the ratio is, do not buy Claude Code capacity; if agent headroom is the goal, the best AI coding tools guide and the Claude Code review are the relevant comparison, not the credit line.

The wider lesson is a procurement one. A subscription whose price is fixed and whose included allowance is adjustable without notice is a variable-cost product sold as a fixed-cost one. The defence is to buy monthly where monthly is offered, to re-check the allowance rather than the price, and to treat any vendor figure expressed as a multiple of another plan — “5x”, “10x”, “20x” — as a number that can move while the plan name stays still. On this evidence, it can also move while the plan name moves instead.

Questions or corrections? Email Pick Right. Want the full list? See all news.