Claude Ultra replaces Max 20x: no 5-hour limit for verified accounts, and Anthropic says the weekly 10x never changed
The 20x in the old name described the 5-hour window, the friendlier of two meters. By Anthropic's account the weekly cap was 10x Pro from the day it was added. Ultra's pricing now leads with the weekly number and drops the 5-hour one, which is a better deal for some people and a correction for anyone who read 20x as weekly volume.
Contents
Anthropic has renamed Max 20x to Ultra. The price stays at $200 a month, existing subscribers moved over automatically, and the plan now promises “10 times the Pro plan’s usage every week,” with no 5-hour limit on verified accounts, per the Help Center and the pricing page, both read on 1 November 2026. Max 5x is now just Max, $100 a month.
The quick reading is that 20x became 10x, so the top plan was halved. Anthropic’s answer is that nothing was halved: the 20x never described the week. Mike Overell, who leads product for Claude subscriptions, wrote on 30 October that the plan launched in April 2025 with a single 5-hour limit at 20 times Pro, that the weekly limits added later were set at 10 times Pro, and that Ultra keeps that weekly limit and drops the 5-hour one. He apologized for the confusion. Taken at his word, nobody lost anything on 30 October. The name caught up with a meter that had been there for months.
What changed in Claude’s Max and Ultra plans?
Two plans, and now one unit for both:
| Max ($100, was Max 5x) | Ultra ($200, was Max 20x) | |
|---|---|---|
| 5-hour limit | Yes, 5x Pro | None on verified accounts |
| Weekly limit | 5x Pro | 10x Pro |
| Fable | Up to 50% of the weekly limit | Same |
| Monthly API credits | $100 | $200 |
Fable is a sublimit. On both plans it draws on the weekly allowance and can use at most half of it, per the Fable plans article. Past that, usage credits or another model. The weekly allowance resets at a fixed time assigned to your account, as the Max article puts it.
Was Claude Max 20x ever 20x Pro per week?
Not by Anthropic’s account. When I wrote up what Claude Code costs on each plan from the pricing page and the Help Center on 24 October, the multiples I recorded were per session: five and twenty times Pro, a session resetting every five hours, weekly limits on top. That write-up has no weekly multiple. Overell’s post now supplies it, 10x, and says it has been 10x since weekly limits were introduced.
So the old $200 plan carried two meters with two different multiples, and the name advertised the generous one. For someone who ran into the 5-hour window and finished the week with room left, 20x was the number that described their experience. For anyone whose weekly allowance ran out before the session one, the number that mattered was 10x, by Overell’s account, and the plan’s name never said so.
One thing this doesn’t settle is size. Anthropic still publishes no fixed token allowance for Pro, so 5x and 10x are ratios to an unknown. I made the same complaint about Notion’s AI pricing: the fee is known, the quantity it buys is not.
Is Claude Ultra worth $100 more than Max?
On the weekly meter, Ultra is 10x Pro and Max is 5x: twice the allowance for twice the price, so the same weekly usage per dollar. What the extra $100 buys is the doubled weekly allowance itself, the missing 5-hour window, and another $100 of API credit, which has to be claimed against a Console organization and does not top up subscription usage.
Bursty work gains the most. A long agent run, or several Claude Code sessions in parallel, could stop at the session cap with most of the week unused. On a verified Ultra account that wall is gone. Steady work gains less, and some of the risk moves. The 5-hour window was an annoying meter, but it also paced you. Without it, a heavy Monday and Tuesday can spend most of the week’s allowance before anything pushes back. Pacing is now your job, which makes rate_limits. in Claude Code’s statusline the number to watch. The docs, as of 1 November, list Ultra among the plans that receive rate_limits and say each window “may be independently absent,” so a script that assumes five_hour exists should handle it missing.
How do you know your Claude Ultra account is verified?
The Help Center’s test: open Settings, then Usage. No “Current session” bar means the account is verified and has no 5-hour limit. Unverified accounts get a temporary one. The Help Center names account and payment history among the inputs, says verification is automatic, and says a phone number verified at checkout does not remove the limit. It gives no timeline. A new subscriber can pay $200 and still hit a 5-hour wall for a while.
Claude Ultra and OpenAI Pro 200: a rename and a canceled reduction
OpenAI had its own 20-to-10 story in October. As I noted in the Notion and ChatGPT token-sharing piece, the Work and Codex allowance of its $200 Pro plan was due to drop from 20 times Plus to 10 times on 30 October for existing subscribers. OpenAI canceled that reduction on 29 October. Its Pro tiers page now says “The reduction announced in our September 29 email is canceled,” and Pro 200 subscribers already on the lower allowance had it doubled to match.
So in the same week, both $200 plans had a 20-to-10 headline, and the two stories went opposite ways. OpenAI withdrew a reduction. Anthropic says there was no reduction to withdraw, only a name that described the other meter. The second is harder to check from outside: among the sources here, the old 10x weekly figure comes only from Overell’s account after the fact.
My read on Claude Ultra
If you were on Max 20x, as I was when I wrote about the programmatic credit in May, the useful question is not whether 10 is less than 20. It is which meter used to stop you. If it was the 5-hour window, Ultra is better at the same price. If it was the week, Anthropic says nothing changed. A log of seven_day at a known load from before the rename is the closest thing to a check on your own account: it can’t tell you the ratio to Pro, but it can tell you whether the same work now uses more of the week. Without one, you can watch the meter from now on and little else.
This outlives the plan. When a price has two meters, the name tends to carry the generous one, and the one that binds is the one you have to go looking for.
01 / Discussion
No comment section here — all discussions happen on X.
02 / More articles
Claude Code worktrees: what claude --worktree isolates, and the ports, databases and installs it doesn't
Tested on Claude Code 2.1.316: two worktrees collided on one port, Docker Compose planned a separate Postgres volume for each, scripted runs left locked checkouts and leftover branches, and the default base skipped the unpushed commit.
Prompt injection examples against coding agents: where the text got in, and what, if anything, stopped it
Nine public cases, sorted by the door the attacker's words came through. In most of them nothing held during the attack. The fixes that came after were all of one kind: a deterministic check re-aimed or a capability removed. A model's judgment that failed had nothing to re-aim.
Vibe coding vs agentic engineering: what still checks the code after the session ends
Simon Willison draws the line at a person who reviewed the code and can explain it. That review happens once, and the code keeps changing after the chat closes. If agentic engineering is a different thing and not a nicer name, the difference is a check that runs without anyone in the room, and that has been seen to fail for the right reason.