Anthropic Drops the 5-Hour Session Limit on Claude Ultra for Verified Accounts
Max 20x Becomes Ultra at $200 as Anthropic Bets on Uninterrupted Agentic Sessions
Anthropic renamed its $200-per-month Max 20x plan to Ultra and, on October 30, 2026, began removing the 5-hour usage limit for verified accounts. The $100 Max 5x tier was simultaneously renamed to simply Max. Both prices are unchanged. The important qualification is that the 5-hour cap is not gone for everyone: newer accounts keep a temporary session limit until Anthropic verifies them, a step the company says is designed to prevent fraud and abuse. Once verified, a notification goes out and the restriction lifts.
What the change actually does is not expand total capacity. Weekly limits on Ultra remain at 10 times the Pro weekly allowance — unchanged from what Max 20x offered before. The gain is structural, not volumetric: where a heavy Claude Code session could previously exhaust the rolling 5-hour window and halt mid-task, verified Ultra subscribers now have a single, undivided weekly budget to spend however their workflow demands.
Why a 5-Hour Cap Creates Specific Problems for Agentic Work
The friction the limit imposed was not symmetrical across use cases. A subscriber using Claude conversationally — drafts, research questions, document review — rarely ran into the session ceiling, because typical interactions consume compute in short bursts spread across the day. The cap hit hardest at agentic and developer workflows, particularly Claude Code, where a single task could run for hours.
All Claude surfaces — the web interface, desktop app, mobile, and Claude Code — draw from a single unified usage pool. That means a multi-hour Claude Code debugging run, a long-form research agent, or an extended scaffolding session all burned through the same 5-hour window that a chat session used. When the window ran out, Claude would block new requests and report a reset time, sometimes hours away. In practice, that meant losing the working context of an agent mid-run and having to restart or reschedule.
The weekly budget is a softer constraint for intensive users. If a developer spends 12 hours on a complex codebase in a single day and burns through a large fraction of the weekly allowance, that is a known trade-off. What was untenable was having that session arbitrarily segmented into 5-hour windows, forcing pauses that the work schedule did not require.
Mike Overell, who leads product for Claude subscriptions at Anthropic, framed the change directly in comments reported on October 30: the weekly limit stays, but the session window is being removed so subscribers can keep building without interruption.
What Ultra Actually Includes Beyond the Limit Change
The limit removal is the headline, but it arrives on top of changes already made to the plan. On October 7, 2026, Anthropic added monthly API credits to Max and Team plans, a move now reflected in the Ultra tier: verified Ultra subscribers receive $200 per month in Claude Platform API credits, claimable through the Claude Console.
Those credits pay for programmatic API calls — building apps, running agents via the Claude SDK, processing data — and are distinct from the interactive usage covered by the subscription. They do not apply to interactive Claude Code sessions, which continue to draw from the subscription pool. Ultra's $200 in monthly API credits matches the subscription cost dollar-for-dollar, meaning developers who build on the Claude Platform API receive roughly the full cost of their subscription back in programmable compute.
Fable, Anthropic's most capable model as of October 2026, is available on both Max and Ultra at up to 50 percent of the weekly limit. The ceiling is the same on both plans; Ultra's advantage is that 50 percent of a 10x weekly allowance is twice as much Fable access as 50 percent of a 5x allowance. For subscribers whose primary use is Fable rather than Opus 5.5 or Sonnet 5.5, the plan math favors Ultra on a per-model basis as well as on total volume.
Claude Code's /limit-reset Command and the Max Tier Gap
For subscribers on the $100 Max plan, an undocumented /limit-reset command has appeared in Claude Code for some users. The command clears the 5-hour session cap once per week, giving those Max subscribers a single manual bypass for particularly intensive sessions. It is not a removal of the limit — the cap returns after use — but it provides a pressure valve for the highest-friction interruptions. Anthropic has not included it in its official command reference, and not all accounts report access to it.
That distinction matters for the upgrade decision. Max at $100 remains a meaningful plan for users whose work is spread relatively evenly across the week and who rarely encounter the session ceiling. The /limit-reset command exists specifically for Max users who hit the wall occasionally but not consistently enough to justify doubling their subscription cost.
Competitive Context: OpenAI Pro and the API Credit Race
The Ultra rename and limit removal arrives during active plan restructuring across the leading AI subscriptions. OpenAI briefly paused new sign-ups to its $200 Pro tier on September 10, 2026, amid capacity pressure, then reopened it at DevDay on September 29. At the same event, OpenAI launched a $500 Pro 500 tier with an "Ultrafast" computing layer and removed its own session limits for Pro subscribers.
The two companies are diverging on how they bundle developer access. Anthropic pairs the $200 Ultra subscription with $200 per month in Claude Platform API credits — a separate programmable compute pool for building applications — making it a single billing relationship for both conversational and agentic development. OpenAI's $200 Pro plan does not include an equivalent API credit bundle.
The Unchanged Weekly Budget and What It Actually Means
The Ultra change is felt exclusively by the cohort that wants to run long, continuous, compute-intensive sessions: a developer spending a full day on a refactor, a researcher running an extended agent loop, an analyst processing a large dataset in a single sitting. For users whose work is distributed across shorter interactions throughout the week, the weekly cap remains the only governing constraint — as it has always been.
For the workflows where agentic AI delivers the most value, session continuity is not a convenience. It is the difference between a tool that fits professional work and one that requires managing around an arbitrary clock.