Anthropic Renames Claude Max Plans to Max and Ultra; Ultra Removes 5-Hour Usage Limit

Just moments ago, Anthropic officially announced: a major upgrade to Claude's top-tier paid plans!

The existing Max 5x plan is officially renamed to Max (the official website monthly fee remains unchanged at $100).

The existing top-tier Max 20x plan is officially renamed to Ultra (the official website monthly fee remains unchanged at $200).

But what's truly exciting is the next sentence—

"Starting today, we will remove the 5-hour usage limit for verified Ultra accounts!"

What does this sentence mean? Those who haven't been put in Claude's "penalty box" can hardly appreciate that suffocating feeling.

In the past, even if you paid a hefty sum for the top-tier plan, Anthropic would still use a "reset every 5 hours" quota to choke you off.

You're in the flow writing code, and suddenly you get limited—that feeling is truly crushing.

But now, the shackles have been completely smashed!

Claude Code's father Boris Cherny and Alex Albert both confirmed the news at the first opportunity.

For most Ultra subscribers, this benefit takes effect starting today! Immediately! You can grind all day long and no longer have to watch the countdown while writing code.

Anthropic and OpenAI's quota war has simply heated up to a white-hot level.

Digging into the official pricing page: from commoner to tycoon!

This adjustment starts on the day of the announcement and will cover most existing Ultra subscribers. Newer accounts still have temporary limits, which will be lifted after automatic verification, with an email notification.

It seems that to prevent "freeloaders" from reselling APIs, Anthropic has held something back.

Meanwhile, the $100-per-month Max continues to retain the five-hour cap.

In addition, the total weekly limit remains unchanged. But even so, breaking the 5-hour limit is still a major concession on Claude's commercialization path!

What exactly is the official play here? We rushed to dig through the brand-new pricing page.

Only upon looking did we realize that Anthropic now divides users' "classes" so clearly, with knife skills comparable to Cook.

In their eyes, users are divided into the following tiers.

Tier 1: Freeloaders and light users

Free (free version): Experience Claude 3.5 Sonnet at zero cost, but the quota is pitifully small.

Pro ($17/month): The most familiar standard for working professionals. More usage than the free version, support for Projects/Claude Code and other features, and access to Claude's core productivity.

Tier 2: Hardcore developers (the core of this upgrade)

Max ($100/month, formerly Max 5x): Heavy users who find the Pro version completely insufficient, you're in luck. At this tier, your compute quota is greatly increased, with extremely high cost-effectiveness.

Ultra ($200/month, formerly Max 20x): This is today's protagonist! The distinguished Ultra club! Those who can buy this plan are basically seasoned veterans.

In addition, the developer plan includes all the benefits of the individual plan plus more additional benefits, but the price remains unchanged!

Anthropic's logic in removing Ultra's 5-hour limit this time is very clear—

Since the big shots are willing to pay $200, then we must provide true "first-class service."

Here comes the hidden Easter egg

Now, Anthropic has done a comprehensive review of the benefits ledger for all paid plans.

However, after the revamp, some users discovered that after Max 20x was renamed Ultra, the official website actually says "weekly quota is 10x that of Pro"?

On the left is the new official pricing page, which says "10x Pro weekly usage"; on the right is the pop-up users see when switching plans in their account backend, which still displays the old "20x Pro usage" copy.

Could it be that Anthropic is baiting and switching, secretly cutting quotas?

Take a careful look at the documentation, and it becomes clear.

First, let's clarify what the 20x actually means.

The old Max 20x's advertised "20x" only measured how many times the Pro user's message volume you could fit within a single 5-hour window. But stretched over 7 days, the $200 tier's total weekly cap has always been fixed at 10x that of Pro.

After Ultra removed the 5-hour single-window limit, "20x per single window" lost its reference point, so the official website simply switched uniformly to "total weekly cap is 10x that of Pro." As for the 20x wording still remaining in the backend plan-switching pop-up, that's purely because the old pop-up was missed during the frontend update.

Heavy users should note a hidden rule.

If you frequently call the top-tier Fable model on the Ultra plan, it has an independent weekly defense line: you can use at most 50% of your total weekly usage, which works out to 5x Pro's weekly quota.

Aside from the total weekly quota being left intact, Anthropic has also stuffed a real, tangible Easter egg for Ultra subscribers this time: the monthly free Claude platform (formerly Anthropic Console) API credit has doubled from the Max tier's $100 to $200!

This $200 API credit is automatically deposited each monthly billing cycle and zeroed out at month's end. Whether you use it to call APIs and run automation scripts, or to experience Claude Managed Agents hosting services, it's all essentially free money.

Because the plan arithmetic from the AI giants has been so convoluted these past two days, even Netflix platform head David Leen joked that you'd probably need Lean to figure out these discounts.

Just yesterday, OpenAI officially announced: the quota reduction is canceled, and all ChatGPT Pro tiers' quotas are doubled.

Pro 100, Pro 200, Pro 500—these three tycoon tiers—as well as Business Premium, all have their usage caps doubled, effective immediately!

The originally planned quota reduction for Pro 200 was also swiftly taken down.

The reason was that OpenAI previously felt compute couldn't keep up, and originally planned to quietly cut Pro 200's quota, taking a knife to Work and Codex usage.

As soon as the news broke, the entire community exploded: "Refund! Refund immediately!"

This explosion scared OpenAI's management straight, and quotas went from halved to doubled in just one night.

Tibo has been temporarily crowned "Saint T"

Why are both companies now fixated on these expensive $100 and $200 plans?

Clearly, the $20/month Plus and Pro users are no longer attractive to large model companies.

Now they're competing for high-net-worth customers—professional programmers, analysts, and architects.

Most heavy CC users are not price-sensitive; what they want is smoothness and stability.

Removing the 5-hour limit is a major boon.

And for OpenAI's Pro 200 and Pro 500 users, if they were driven away over a few dollars of compute cost, it would be a huge blow to OpenAI's ecosystem.

Today you remove limits, tomorrow I double quotas—this precisely reflects the truth—

The gap between underlying models is narrowing, and compute quotas and user experience are becoming the real moat.

References:

This article comes from the WeChat public account "Xinzhiyuan" (ID: AI_era), author: ASI Revelation, published by 36Kr with authorization.