Claude Ultra Replaces Max 20x: No More 5-Hour Limits

What Anthropic Actually Changed on October 30

Anthropic shook up its consumer subscription ladder on Friday, October 30, announcing via its official X account that the two top Claude plans are getting new names — and one of them is losing its most hated restriction. Max 5x is now simply "Max" at $100/month, and Max 20x is now "Ultra" at $200/month. Prices didn't move, and existing subscribers were migrated automatically: no re-subscribing, no prorating, no action required.

The headline change is bigger than the rebrand. Verified Ultra accounts no longer face the rolling five-hour usage cap that has defined the Claude power-user experience since the Max plans launched. Anthropic said the removal applies "starting today with most existing subscribers," with the five-hour window surviving only as a temporary limit for accounts that haven't completed verification.

The rename also quietly retires a source of confusion. The "20x" label led plenty of new subscribers to assume they were getting 20 times a Pro plan's usage — they weren't. Anthropic's subscriptions product lead Mike Overell publicly acknowledged the mismatch and apologized for it. The new names describe tiers, not multipliers, which is more honest even if it's less specific.

Inside the Ultra Plan: What $200/Month Gets You

Per Anthropic's pricing page and help center, the Ultra plan now includes:

• 10x the Pro plan's usage every week — a weekly allowance that applies across all models, with a separate weekly limit for the Fable model family (up to half your weekly limit)
• No five-hour session limit for verified accounts — you "work at your own pace within your weekly limit"
• $200/month in Claude Platform API credits — doubled from the $100 included with Max, claimable against the Claude API, the Console Playground, Claude Managed Agents, and the Agent SDK
• Claude Code included, drawing from the same usage pool as your web, desktop, and mobile chats
• Priority access to new models, features, and products, plus priority at high-traffic times
• Longer background tasks — hand Claude reports, spreadsheets, and presentations and it keeps working unattended

One billing note: Ultra is monthly-only. There's no annual discount, and the API credits don't roll over — unused credit expires at the end of each cycle. The credits also don't cover interactive Claude Code sessions or overage "usage credits" inside the chat product; those are billed separately on top of the subscription.

The Catch: The Weekly Cap Didn't Go Anywhere

Anthropic was explicit that "weekly limits remain unchanged." Removing the five-hour window doesn't add usage — it removes the forcing function that spread your allowance across the week. A subscriber who previously couldn't burn through their full weekly allowance because the session cap kept pausing them can now exhaust it faster. Once the weekly pool is empty, you either wait for the reset or switch on usage credits billed at standard API rates.

And the standard rates are steep at the frontier: Anthropic lists Opus 5.5 at $4 input / $20 output per million tokens, and Fable 5.1 at $10 / $50. Heavy Claude Code users who blow past the weekly cap can rack up meaningful overage bills quickly — the same reason the Max plans felt like a bargain in the first place (community math has pegged a Max 20x subscription as worth thousands of dollars at API rates) works in reverse once you're paying per-token above the cap.

Anthropic also reserves the right to apply "additional usage limits to manage capacity," including weekly and monthly caps and per-model limits at its discretion. No Claude plan publishes a message count, so the only reliable way to know where you stand is the in-product usage indicator.

Max vs Ultra: Who Should Actually Upgrade

The decision now hinges on a single question: what stops you first — the session window or the weekly pool?

• Stay on Max ($100/month) if you use Claude throughout the day but rarely hit the five-hour wall, or don't mind pausing when you do. You get 5x Pro weekly usage, the $100 API credit, and the same priority access for half the price.
• Upgrade to Ultra ($200/month) if your work runs in long, unbroken stretches — marathon Claude Code sessions, multi-hour background agents, weekend refactors — where the session cap is the interruption that hurts. Ultra doubles your weekly allowance to 10x Pro and doubles the API credit.

The removal of the five-hour limit is most valuable to exactly the people who were already frustrated enough to consider leaving: agent-heavy users running long autonomous tasks, where a session reset mid-task isn't an inconvenience, it's lost work.

How Ultra Fits the 2026 Coding Subscription Landscape

Claude Ultra doesn't exist in a vacuum. If you're deciding where $200/month of AI budget goes, the honest comparison set includes:

• Claude + Claude Code — Ultra's pitch is one subscription covering chat, the terminal agent, and API building. Claude Code remains the reference terminal coding agent for long-horizon refactors.
• ChatGPT and OpenAI's Codex line — OpenAI's own high-tier subscriptions bundle frontier models with agent tooling, and GPT-6.1 Terra's computer-use push is aimed squarely at the same all-day agent workload.
• Cursor and Windsurf — IDE-native agents that bill per-partial-seat rather than per-power-user, often cheaper for teams that don't need unlimited frontier-model chat.
• GitHub Copilot — the enterprise default at roughly $10–39/seat, with agent mode closing the capability gap for many everyday tasks.
• Gemini CLI — Google's terminal agent, frequently the value play for developers with generous free tiers and fast model access.

The practical pattern in late 2026 is stacking, not loyalty: a mid-tier IDE subscription for flow-state edits, plus one frontier subscription — Ultra, a ChatGPT top tier, or Gemini's equivalent — reserved for the hardest work. If Claude is your hardest-work model, Ultra's no-session-cap model plus $200 in API credits makes it the most flexible of the frontier bundles for developers who ship against the API as well as chat.

Frequently Asked Questions