Claude Free vs Pro vs Max vs Ultra: What Each Plan Actually Gives You

Claude now comes in four individual flavours — Free, Pro, Max and Ultra — and the names tell you almost nothing about what separates them. The real dividing line is not features; it is usage: how much Claude you may use before the meter stops you, and how the meter works. This guide walks through each tier using Anthropic’s own official pricing page and help centre, both checked on 7 November 2026, and explains the five-hour and weekly limits that decide which tier a heavy user actually needs. It is part of our complete guide to AI subscriptions and pricing explained. Prices exclude tax and, as Anthropic notes, can change at its discretion — always confirm on the pricing page before you buy.

Free ($0): the full product, on a short leash

Anthropic’s free tier is not a demo. It includes chat on web, desktop and mobile, web search, the ability to create files and run code, memory across conversations, connectors, Artifacts and up to five Projects, with model training opt-out available on every tier. What you do not get is volume: usage is limited both within each five-hour window and across each week, and at busy times free users queue behind paying ones. For everyday questions, drafts and the occasional long document, Free is genuinely sufficient — Anthropic positions it for light, everyday use. The moment Claude becomes part of your working day, the limits, not the features, push you upward.

Pro ($20/month, or $17/month billed annually): the working default

Pro is where Claude becomes a tool rather than a treat. Anthropic guarantees at least five times the Free plan’s usage per five-hour session, and adds the capabilities professionals actually ask for: Claude Code for terminal-based coding, full Projects, access to more models including the flagship Opus line, Claude inside Chrome and Microsoft 365, and the ability to hand off and schedule tasks. Billing is the friendly kind: $20 month to month, or $17 a month if you commit to a year — $200 billed up front. For most writers, analysts, students and developers who use AI daily but not constantly, Pro is the tier the range is designed around, and the comparison in our ChatGPT vs Claude vs Gemini test shows why its limits rarely bite casual professionals.

Max ($100/month) and Ultra ($200/month): buying the ceiling away

Max and Ultra exist for people who hit Pro’s ceiling so often that the interruptions cost more than the upgrade. Anthropic’s help centre states the core fact plainly: the Max plan includes five times the Pro plan’s usage every five hours and every week, and costs $100 per month, billed monthly only. Max also bundles $100 a month in API credits, lets you spend up to half your weekly limit on the top Fable models, and adds early and priority access at busy times. Ultra, at $200 a month, doubles down: ten times Pro’s weekly usage, $200 a month in API credits — and, for verified accounts, no five-hour session limit at all, leaving only the weekly meter running. That last detail matters more than the multipliers: it removes the mid-afternoon wall that interrupts long coding or writing sessions.

Notice what Max and Ultra do not add: a smarter Claude. The models, the context window (up to 1M tokens, varying by model) and the feature list are essentially the Pro product. You are paying for throughput and priority — which is exactly right for professionals billing by the hour, and exactly wrong for anyone who simply wants “better answers”.

How the limits actually work

Anthropic meters usage in two overlapping windows. The session limit is a rolling five-hour window: work draws it down, and it refills five hours after your session began. The weekly limit, on paid plans, sits on top — a heavy Monday can leave you short on Friday even though every session in between reset. Crucially, all surfaces share one pool: web, desktop, mobile and Claude Code draw from the same allowance, so a morning of coding spends the same budget as an afternoon of chat. You can watch both meters in Settings > Usage, where Anthropic also shows your next reset time; the weekly reset falls at a fixed time assigned to your account. When you hit a limit, Anthropic gives you three exits: wait for the reset, move up a plan, or — on paid plans — switch on usage credits and pay standard API rates for the overflow. If you are weighing Pro against a rival before committing, our side-by-side of ChatGPT Free vs Plus shows the competing meter, which is counted in messages and feature caps rather than windows.

Which tier should you actually buy?

Start on Free for a fortnight and watch how often you are stopped. Stopped rarely? Stay free — the product is the same. Stopped most days, or you need Claude Code and Projects for real work? Pro, billed annually if you are confident you will stay. Hitting Pro’s weekly limit regularly, or losing billable time to resets? That is Max’s exact use case — and do the arithmetic honestly: at professional rates, one saved hour a month can cover the difference. Ultra is a narrow product: verified heavy users, typically developers running Claude Code all day, for whom the five-hour wall itself is the problem. Everyone else is paying for headroom they will never use. And before any upgrade, read our guide to AI subscription traps — Anthropic’s own terms note that most Claude payments are generally non-refundable, and cancelling needs to happen at least 24 hours before renewal to avoid the next charge, so set the reminder the day you subscribe.

Side by side

What this means

Claude’s pricing is the clearest example in the industry of what AI subscriptions have become: not software licences, but metered capacity. The four tiers sell one product at four flow rates, and the honest question they pose is not “which Claude do I want?” but “how many hours a week do I actually spend with it?” That reframing cuts both ways. It protects light users — Free and Pro are fairly priced for what they include — and it quietly taxes the heaviest users, whose $200 Ultra bill buys the same answers a free user gets, just without the waiting. Compared with rivals, Anthropic is unusually transparent: multipliers instead of vague “generous limits”, a visible usage meter, and overflow priced at published API rates rather than a forced upgrade.

For US readers, the practical effect is a budgeting shift. A single Pro subscription at $17–20 a month is now a normal professional expense, in the same mental category as a music or storage plan. Max and Ultra, though, belong in a different ledger: they are business inputs, justified only when the hours they unlock are billed to someone else or replace a more expensive tool. Teams should also note that Anthropic sells separate Team and Enterprise seats rather than expecting staff to stack personal Max plans. The healthiest habit is quarterly: check Settings > Usage, compare it with what you paid, and let the meter — not the marketing names — choose your tier. In a market where every assistant claims to be “Pro”, Claude’s range at least tells you the truth about what the money buys: time.

One boundary in the range is worth watching: the API credits bundled with Max and Ultra blur the line between a subscription and raw API access. Heavy users should price both routes before renewing — a professional whose work is mostly scripted or tool-driven may find metered API billing cheaper than a $200 tier, while a chat-and-coding user gets better value from the subscription’s predictable pool. Anthropic is, in effect, selling the same compute through two doors. Knowing which door you are standing at is the whole skill this pricing model demands.

Bottom line

Free is a real product for light, everyday use, and Pro at $20 a month is the working default once Claude becomes part of your day. Max and Ultra only make sense when you regularly hit Pro’s ceiling — what you are buying is headroom, not features, so let your own limit messages, not the tier names, tell you when to climb.

Read Next

• AI Subscriptions and Pricing Explained: The Complete Guide
• Gemini Free vs Google AI Plus, Pro and Ultra: The Plans in Plain English

Sources

• Anthropic — Claude Pricing (plans, usage limits and billing FAQ) (accessed 7 November 2026)
• Anthropic Help Center — What is the Max plan? (accessed 7 November 2026)