Claude Ultra: Anthropic Drops the 5-Hour Limit on Its $200 Plan

Anthropic's pricing page now lists Claude Ultra as a $200 per month individual plan with no 5-hour limit for verified accounts, 10x Pro usage every week, and $200 a month in API credits (claude.com/pricing). Max stays at $100 per month. The names replace the older "Max 5x" and "Max 20x" labels, which had confused people about what the multiplier actually applied to, according to reports of statements by Anthony Overell, who leads product for Claude subscriptions.

If you pay for Claude to run long Claude Code sessions, this is a pricing and limits change worth reading line by line. The headline is generous. The detail is that the weekly cap is still there.

TL;DR: Free, Pro, Max and Ultra side by side

All figures below are from Anthropic's pricing page on October 30, 2026.

What exactly changed

Three things, per the pricing page and its FAQ:

1. A rename. The $100 tier is now simply Max, and the $200 tier is Ultra. Prices did not change.
2. The 5-hour limit is gone on Ultra. The page says "No 5-hour limit," with a footnote that it applies to verified accounts. Every other plan still resets on a rolling five-hour session window.
3. API credits scale with the tier. Ultra includes $200 a month of Claude Platform credit, up from $100 on Max, and the FAQ says both Max and Ultra include Fable for up to half your weekly limit.

Reports of statements from the Claude account and from Overell say the 5-hour limit is being removed in stages, starting with most existing subscribers, to prevent fraud and abuse, and that the weekly limit stays (Overell is quoted as describing it as 10x Pro). We could not open those posts directly, so treat that attribution as reported, and rely on the pricing page for the numbers.

Why the old names caused confusion

When Anthropic launched Max in April 2025, it sold a $100 tier with five times the usage of Pro and a $200 tier with twenty times (Engadget's launch coverage). Many users read "20x" as a multiplier on everything. The reported explanation for the rename is that the label applied to the session cap, not to every limit.

Look at the new numbers closely: Ultra's weekly limit is listed as 10x Pro, while Max is 5x Pro. So Ultra is twice Max on weekly usage, and removes the session window entirely. We have not verified how this compares with the weekly allowance the old 20x tier had, and the pricing page does not say. If you are a heavy user on the old $200 tier, compare your own weekly usage meter (Settings, then Usage) before and after the rollout rather than assuming parity.

Who this actually helps

A balance scale weighing a subscription card against token chips, showing the Claude Ultra plan trade-off

The 5-hour window is the limit most developers hit mid-task. Claude's own FAQ says all usage across web, desktop, mobile and Claude Code draws from one pool and resets on a rolling five-hour window, with weekly limits layered on top. Removing the window on Ultra helps three kinds of user:

• Long agentic runs. A multi-hour refactor or background task no longer stalls when the session window fills. Our token efficiency guide explains why long sessions burn the pool faster than people expect.
• Parallel sessions. Several Claude Code terminals share one pool, so bursts hit the window quickly.
• Anyone who builds on the API too. The $200 credit is a real offset if you already pay for Claude Platform usage.

It helps less if your problem is the weekly cap. A steady heavy week can still run dry, and Anthropic's FAQ is explicit that it "may limit your usage in other ways, such as weekly and monthly caps or model and feature usage, at our discretion."

Is Ultra worth $200? A simple way to decide

The credit makes the arithmetic easier. If you would otherwise spend at least $100 more a month on API usage, the extra $100 of included credit alone covers the step from Max to Ultra, and the session-window removal and doubled weekly allowance are on top. If you never touch the API, the question is only whether you hit the 5-hour wall regularly.

A practical test: for one week on Max, note how many times you were blocked by the session limit and how close you got to the weekly meter. Blocked more than a couple of times and still below the weekly line is the profile Ultra is built for. Hitting the weekly line, not the window, means Ultra only doubles your ceiling.

What people are asking

Is Ultra the same product as the old Max 20x?

The prices match ($200) and the tier sits in the same place on the ladder, so the rename looks like a relabel plus new benefits. But the page lists the weekly limit as 10x Pro and adds the missing 5-hour cap and the larger API credit, so it is not a pure rename. We could not confirm a before-and-after weekly figure from Anthropic.

Do I have to do anything to get the change?

The pricing page does not describe any opt-in. Reports say the 5-hour limit is being lifted for existing verified subscribers starting October 30, with a staged rollout. If your meter still shows a session window in Settings, the change has probably not reached your account yet.

Can I still use usage credits?

Yes. The FAQ says that when you hit a limit on a paid plan you can wait for the reset, move to a higher plan, or turn on usage credits to keep working at standard API rates. That matters for Ultra users who run into the weekly cap: credits are the escape hatch, billed at the model prices Anthropic lists, which on the same page are $4 input and $20 output per million tokens for Opus 5.5 and $10 input and $50 output for Fable 5.1.

Does the Team plan get any of this?

Not directly. Team standard seats give more usage than Pro and premium seats give 5x more than standard, at $100 per seat per month on annual billing. Premium seats also include Fable for up to half of weekly limits, and the Team plan lists up to $500 a month in pooled API credits. The Ultra change is on the individual page only.

How to get more out of a plan with no session window

Removing the window changes how you should work, because the only meter that matters is the weekly one. A few habits follow from that:

• Watch the weekly meter, not the session reset. Settings, then Usage, shows where you stand. Long unattended runs can eat a week faster than a day of normal chat.
• Pick the model for the job. Fable counts against the weekly pool and can use up to half of it. Using Sonnet or Opus for routine edits keeps the heavy model for the work that needs it.
• Keep sessions focused. Long conversations carry the whole history on every turn; starting a fresh session for a new task is cheaper than extending a huge one.
• Cut waste at the harness. Unused tool definitions and repeated failing loops cost tokens whatever your plan is, which is the problem a plugin like TokenEconomizer targets.

These are general practices drawn from how Anthropic describes its limits, not rules the pricing page states.

How this fits the wider limits fight

Limits are the main way labs ration scarce compute, and they keep moving. We covered Anthropic's earlier rate-limit boost and the developer debate against Codex, and OpenAI went the other direction when Codex Plus got its 5-hour limit back. On the enterprise side, Accenture capped central Claude funding at a per-person amount, a sign that organizations are moving toward consumption budgets while individuals are being offered flat tiers.

If you want to stretch any plan further, cost control at the agent layer helps: Microsoft and Berkeley's TokenEconomizer plugin reported 18.4–30.9% lower SWE-Bench cost on Claude Code, Codex and Copilot CLI. And if you route around plan limits with third-party proxies or account switching, read the terms-of-service and data risks first, as in our look at the free Claude Code proxy.

What is still unclear

• Verification criteria. "Verified accounts" is not defined on the pricing page.
• Rollout timing. Reports describe a staged removal; the page does not give dates.
• Existing 20x subscribers. Whether anyone's weekly allowance went down in the rename is not stated.
• Fable access. The page says Fable is included up to half the weekly limit on Max and Ultra and is paid with usage credits on Pro, so heavy Fable users will still draw down the weekly pool quickly.

We will update this post if Anthropic publishes the verification rules or the weekly numbers in more detail.

Related reading

• Claude Code token efficiency and prompt-cache guide
• Claude Code vs Codex after Anthropic's rate-limit boost
• Codex Plus gets the 5-hour limit back
• Accenture caps central Claude funding
• TokenEconomizer for Claude Code, Codex and Copilot CLI
• Free Claude Code proxy: terms and data risks

Plan names, prices and limits are as shown on Anthropic's pricing page on October 30, 2026, and are subject to change at Anthropic's discretion.