Claude Max vs Ultra: $100 or $200 a month, and the limit that should decide it

Claude Max vs Ultra comes down to which limit stops you. Max costs $100 a month for 5x Pro usage in every five-hour window and every week. Ultra costs $200 for 10x Pro usage per week and, on verified accounts, no five-hour limit at all. If the five-hour reset is what interrupts your Claude Code sessions, Ultra fixes that. If you rarely hit it, you are paying mostly for a bigger weekly cap and $100 more in API credits.

By the end of this post you will know what each plan includes as of November 2026, what the API credits can and cannot pay for, and a simple test for which plan matches how you work. Every figure below comes from Anthropic's pricing page or its help center, checked on November 5, 2026.

Claude Max vs Ultra at a glance

The two plans share most features. Both include Claude Code, priority access at high traffic times, and Fable models up to half of the weekly limit. The differences are in the limits and the credits.

Prices are the web prices on the pricing page and exclude tax. Anthropic's Ultra help article notes that mobile pricing may vary, and the pricing page says plans are subject to change.

Claude Ultra is the old Max 20x plan, renamed

The Claude Ultra plan is not a brand-new tier. Anthropic's help article says it plainly: "The Max 20x plan is now called the Ultra plan." The same article says existing Max 20x subscribers move to Ultra automatically with no price change, so if you were paying $200 for Max 20x, you are now on Ultra and do not need to do anything.

The $100 plan is now simply called Max, with 5x Pro usage. The Anthropic help pages we read do not explain the rename, and we have not found an official statement of what the old Max 20x weekly allowance was, so we are not going to compare Ultra's 10x weekly figure against it. What you can compare is today's two plans, side by side, on Anthropic's own pages.

Two things to know before you subscribe. The Ultra plan is available as a monthly subscription only, and if you are moving from an annual Pro plan with a remaining balance larger than the Ultra price, you receive a credit, provided your billing address matches.

The five-hour limit is the real difference

Claude's five-hour limit is a rolling session window. According to Anthropic's usage limits article, the period starts when you send your first message and resets in full five hours later. If you hit it before then, you wait for the reset, even when plenty of the week's allowance is left. For long Claude Code sessions, that wait is the interruption that matters.

Max keeps that window, at five times Pro's size. Ultra removes it, but only for verified accounts. The same help article states that "Verified Ultra accounts don't have a five-hour limit."

Claude Ultra's no 5-hour limit applies after verification

Verification is automatic. The Ultra help article says Anthropic verifies Ultra accounts "using factors like your account and payment history," and you do not need to take any action. Phone verification at checkout is a separate step and does not remove the limit. Until your account is verified, a temporary limit applies, and the article does not say how long that lasts.

You can check your status in Settings > Usage. Per the same article, if you do not see a Current session bar, your account is verified. If you are upgrading specifically to escape the five-hour reset, check that screen in your first week rather than assuming the limit is already gone.

Weekly limits: Ultra doubles Max, it does not quadruple it

The Claude Ultra weekly limit is 10x Pro, against 5x Pro on Max, per the pricing page. Twice the price buys twice the weekly allowance. The extra value of Ultra is that you can spend that allowance in whatever pattern you like, including one long day.

Claude Max 5x vs Ultra 10x when you count Fable

Both plans include Fable 5 and Fable 5.1. Anthropic's Fable article says you can use up to 50% of your weekly limits on Fable models at no extra cost, but they draw from the same weekly limit and use it faster than other Claude models. On Pro, Fable sits outside the plan and runs on usage credits.

Weekly limits reset at a fixed time assigned to your account, not on the day you subscribed, according to the usage limits article. That article also says Anthropic may adjust limits at its discretion and does not guarantee uninterrupted service, which applies to Ultra as much as to any plan.

These limits are shared across tools. Anthropic's Claude Code article says all activity in Claude and Claude Code counts against the same limits, so chat use during the day reduces what is left for coding at night.

Claude Ultra API credits pay for API work, not your Claude Code sessions

Both plans now come with monthly Claude API credits: $100 on Max and $200 on Ultra, per the pricing page. The API credits help article is precise about where they work and where they do not.

• Covered: the Claude API, the Console Playground, Claude Managed Agents and the Claude Agent SDK, with any available Claude model on the Claude Platform.
• Not covered: interactive Claude Code in the terminal, IDE, desktop or web, usage credits inside Claude, Claude Code or Cowork, and Claude on Amazon Bedrock, Google Cloud Vertex AI or Microsoft Foundry.
• Running claude -p counts against the credits only when you use an API key from your linked Console organization. Signed in with your plan, it uses plan limits.
• Credits do not roll over. Unused credits expire at the end of each billing cycle.
• You link one Console organization, you must be on an eligible plan for seven days before claiming, and everyone with an API key in that organization shares the balance.

In other words, the credits do not stretch your Claude Code allowance. They pay for things you build: a script that calls the API, an agent built on the Agent SDK, a batch job. When they run out, API requests stop unless the organization has its own purchased credits or auto-reload, and the help article says usage is never charged to your Claude plan.

What $100 or $200 of API credits buys at list prices

At the rates on Anthropic's API pricing page, Claude Opus 5.5 costs $4 per million input tokens and $20 per million output tokens. Claude Sonnet 5.5 costs $2 and $10. So $100 of credits covers about 5 million Opus 5.5 output tokens, or 10 million on Sonnet 5.5, before input costs. Ultra's $200 doubles that. Real jobs mix input and output, so treat these as ceilings rather than a forecast.

Is Claude Ultra worth it for Claude Code?

Claude Ultra is worth it if the five-hour reset is what stops your work, or if you already spend at least $100 a month on the Claude API in one Console organization. It is the wrong upgrade if neither is true. Here is how to read your own usage.

1. You hit the five-hour limit on Max but finish most weeks with allowance to spare. This is where Ultra helps most. Removing the window lets you spend a weekly budget twice Max's in fewer, longer sittings.
2. You hit the weekly limit on Max. Ultra's 10x weekly cap is double Max's, so it helps, but compare it with usage credits first (below).
3. You rarely hit either limit on Max. Stay on Max. Ultra would mostly buy credits you may not use.
4. You hit the five-hour limit on Pro. Max at $100 is the next step, with a five-hour window five times larger.

The case for usage credits instead of an upgrade

The strongest argument against Ultra is that you may not need another $100 every month. Anthropic's usage credits article says Pro, Max and Ultra subscribers can turn on usage credits, billed at standard API rates, and they apply to both Claude conversations and Claude Code in the terminal. You set a monthly spending cap and can add auto-reload.

If you go over Max a few times a month, capped usage credits may cost less than the $100 step to Ultra. If you go over most days, the flat price is easier to plan around. The catch is that credits do not remove the five-hour window. The usage limits article says that with usage credits enabled your work continues on credits instead of stopping, so you pay per token for anything past the limit rather than having it included.

When the API credits change the math

Ultra costs $100 a month more than Max and includes $100 a month more in API credits. If you already pay Anthropic at least that much each month for API work in a single Console organization, the extra credits offset the price difference, and the higher limits come at little net cost. If your API spend runs through Bedrock, Vertex AI or Foundry, or is split across several organizations, that offset does not apply, because the credits cover one linked organization on the Claude Platform only.

Upgrading from Claude Max to Ultra: what happens to limits and credits

Upgrading is immediate. The usage limits article says that if you upgrade, your new plan's limits apply right away. On credits, the API credits article says you receive prorated credits right away when you move from Max to Ultra, then $200 each billing cycle after that.

Downgrading stops new credits, but credits you already have stay usable until they expire. Moving from Max or Ultra to a Team plan ends the link, and a Team owner can claim the team's credits after the Team plan has been active for seven days. If you bought through Google Play, the Ultra article says upgrades charge the full new price and unused value converts into extra days.

Pick by the wall you hit, then check after a week

Before you pay for Ultra, spend a week on Max watching Settings > Usage. Note whether the Current session bar or the weekly bar fills first, and how much you spend on the Claude API in your linked organization. That one week will tell you more than any comparison table.

If you are weighing Claude against other coding tools, the AI Coding Tools category lists 355 products on Upfynd as of November 5, 2026, and the Claude listing sits among 2,990 products in AI Assistants. For how other vendors handle the same problem, we have covered ChatGPT subscription API credits and their limits and Cursor's X subscription. And if you build tools for Claude Code users, you can list your product for free.