Anthropic removes the 5-hour cap in Claude Code: Ultra only

Anthropic has made its move in its subscription catalog and, in the process, has touched one of the brakes that most annoyed those who use Claude Code daily. The five-hour rolling window cap disappears for subscribers of the top plan, albeit with nuances worth reading before rushing to change tiers. The company announced the change on Friday and, in the same operation, renamed its two most expensive consumer plans: Max 5x becomes Max and Max 20x becomes Ultra. Prices don't move: $100 and $200 per month respectively, and current subscribers don't have to do anything to migrate.

What the five-hour limit really meant

Until now, Claude Code applied a usage cap within a five-hour rolling window. In practice, a demanding task could exhaust that quota much sooner than the clock suggests: an intense debugging session or a large refactor was enough to hit the wall mid-task. The user was forced to wait for the reset even if they had plenty of weekly allowance left. That was the underlying problem: a short-term limit that interrupted work without the overall compute being exhausted.

With the change, Claude Code users on the $200 plan no longer run into that rolling window, as long as their account is verified and they retain available weekly usage. Boris Cherny, creator of Claude Code, highlighted the news in his own post, which gives an idea of how much the issue worried the development community.

The fine print: verification and weekly cap

Not all Ultra subscribers will see the limit disappear instantly. Anthropic requires account verification which, according to the company, serves to prevent fraud and abuse. The process is activated automatically based on account and payment history; there is no way to request it or speed it up, so some users will have to wait. Anthropic will notify by email when the cap is lifted.

And there is a detail worth keeping in mind: the weekly limit remains intact. Without the five-hour window, a user can consume that weekly quota sooner, because there are no longer forced pauses to slow them down. The pricing page places Ultra at ten times Pro's weekly usage, compared to Max's five times. In addition, conversations with Claude count against the same weekly limit as Claude Code sessions, whether on web, desktop or mobile. When that allowance is exhausted, there remains the option of enabling usage credits at standard API rates, paying separately from the subscription.

Included API credits: the other side of the coin

The pricing page also lists monthly Claude Platform API credits: $100 with Max and $200 with Ultra. That is, each plan includes API credits equivalent to the amount of its monthly subscription. For teams using Claude Code while building applications against the API, this adds one more variable to the equation: not only are they paying for the development tool, but part of that outlay comes back in the form of platform consumption.

What it means for an SMB or an IT team

The quick read is that Anthropic is segmenting its offering more clearly: Max for those who need a moderate volume and Ultra for those who squeeze Claude Code as their main development tool. For an SMB with a small team, the relevant question is not whether the five-hour limit is annoying, but whether the usage pattern justifies the jump from $100 to $200 per month.

• If the team rarely exhausts the weekly quota, eliminating the five-hour cap doesn't add much: the bottleneck is elsewhere and the extra expense isn't justified.
• If interruptions due to the rolling window are frequent and weekly allowance remains, Ultra may offset the outlay by avoiding stoppages mid-refactor or mid-debugging session.
• If the weekly consumption is already exhausted before the week ends, removing the session cap won't give more included usage: you'll have to rely on API credits or rethink the plan.

Our recommendation is to measure before deciding. Logging over two or three weeks how many times you hit the five-hour limit and how many times the weekly one gives a much more reliable picture than subjective impression. It's the same logic we apply when we talk about measuring the real performance of AI and not stopping at the superficial indicator: a GPU at 100% doesn't say whether the work is progressing, just as a session limit doesn't say whether the team is making the most of the tool.

Market context: pressure on limits

Anthropic's move doesn't happen in a vacuum. A day earlier, Thibault Sottiaux, of OpenAI, announced higher usage limits for Pro and Business Premium subscribers. The company also reversed a planned cut for existing subscribers of the $200 Pro plan after customer complaints, and abandoned the idea of replacing token-based billing with flat per-task fees after developer protests. The message is clear: providers are competing to retain technical teams, and usage limits have become a commercial battleground.

For an SMB, this is good news in the medium term: there is room to negotiate and to review subscriptions periodically. But it also demands discipline. We recommend reviewing each quarter which plan fits actual usage, keeping a record of consumption per tool and not taking for granted that current conditions will hold. And, as always, beware of unofficial apps that promise access to these models: we've already seen how fake ChatGPT and Claude apps install miners, a risk worth keeping in mind when expanding the company's fleet of AI tools.

In short: Anthropic removes a real friction for intensive Claude Code users, but it does so only on the most expensive plan, with prior verification and without touching the weekly ceiling. Before upgrading your plan, measure, compare and decide with data.

Source: The New Stack. Analysis and adaptation: ForgeNEX.