Claude Pro pricing: what $20 a month gets you

What Claude Pro costs

Claude Pro costs $20 per month billed monthly, or $17 per month if you pay for a year up front. Those are the figures on Anthropic's pricing page as of 9 November 2026, when this page was last checked. Prices vary by region, and annual billing is taken as one payment rather than twelve. Anthropic's Pro help article says the plan is priced in your local currency where supported, and that some regions include tax in the displayed price while others add it at checkout. Brazil is the clearest case of that gap: Anthropic launched Pro there in August 2024 at R$ 110 a month, priced in reais rather than dollars, so what Claude Pro costs you from Brazil depends on the currency your checkout shows and whether tax is already in that figure. The billing currency comes from your region rather than a setting you pick, and changing the currency Claude bills you in means cancelling at the end of the current term and resubscribing, with separate rules for web, iOS and Google Play subscriptions. If your card is refused outright rather than just charged more, why a Claude payment gets declined in an unsupported country explains the cause and the open model you can still run without a subscription. If you are already paying for ChatGPT and deciding whether to run both, the two subscription ladders set side by side shows where each one wins on price for coding work. Pro sits in the middle of a longer range, and every Claude plan priced in one table shows what the tiers either side of it cost and which of them include Claude Code.

The annual option is a single charge of $200 for the year against $240 paid month by month. You save $40 and you commit for twelve months. The commitment changes more than the headline figure, so read how annual billing handles mid-cycle upgrades, refunds and invoices before you hand over a year in one payment. Pay monthly for the first cycle or two, watch what you actually use, then switch if the plan fits. If you change your mind later, cancelling or downgrading leaves you the rest of the period you have already paid for.

The price does not include API access. Pro covers Claude on the web, in the desktop and mobile apps, and in Claude Code. Calling the Claude API from your own code is billed separately per token through the Console, against its own balance. Anthropic offers new users a small credit to test the API, and a few parts of the API cost nothing. Comparing the subscription against API pricing does that arithmetic properly.

What Pro adds over the free plan

The free plan is a real product, not a trial. It gives you chat on web, mobile and desktop, web search, file creation with code execution, memory across conversations, connections to your apps and tools, and Artifacts. What the free tier actually allows covers where it stops. How often it stops you is the thing worth counting, and working out whether Pro is worth paying for turns that count into a decision.

Pro adds five things worth paying for.

• More usage. Anthropic's pricing page says Pro gives you at least five times more usage per five-hour session than Free, plus more usage every week. The Pro help article puts it only as more usage per session than the Free plan, without the number. The other multiples Anthropic publishes are Max's, at five times Pro every five hours and every week, and Ultra's, at ten times Pro every week.
• Claude Code. The terminal agent starts at Pro, and for most developers it is the whole reason to subscribe.
• More model choice. The pricing page lists more Claude models for Pro, including Opus, so you can match the model to the task instead of taking the default. Fable is the exception: on Pro it is not part of the included usage and runs only on usage credits. Picking between Opus, Sonnet and Haiku explains when the cheaper model is the correct one.
• Projects without the Free plan's cap of five, and access to Research, both listed in the pricing comparison table.
• Priority access when traffic is heavy, plus early access to new features. Anthropic's Pro help article still lists both as of 9 November 2026, though the comparison table on the pricing page now marks priority access for Max and Ultra only.

The pricing page also lists Claude Science and Claude Dashboards (beta) under Pro, along with handing off and scheduling tasks, Claude in Chrome and Microsoft 365. Claude Design, Slides and Docs now appear on the Free plan too, so they are no longer a reason to upgrade. On 16 September 2026 Anthropic announced that Cowork and chat are merging into one Claude, rolling out to Pro, Max and Ultra first, with more plans to follow.

Claude Code on Pro, and the shared limit pool

Claude Code runs in a terminal. You install it, then log in with the same account you use on claude.ai. There is no separate purchase and no API key to paste. If you are still working out whether your own plan covers the terminal agent at all, which plans include Claude Code runs down the tiers and the case where a stray API key quietly bills you instead.

Here is the mechanic that surprises people. Pro's usage limits are shared across Claude and Claude Code. Anthropic states that all activity in both tools counts against the same limits, so a long refactor in the terminal spends the same allowance as a long chat in the browser. There is no second quota waiting behind the first. What Claude Code actually costs to run weighs that against paying per token.

Run /status inside Claude Code to see what is left. Settings > Usage on the web shows the same picture plus your next reset time.

The subscription follows your account, not your machine, so running Claude Code on a VPS inside tmux costs exactly what running it on your laptop costs. One warning goes with that. A detached session keeps working while you are away from the keyboard, which makes it easy to spend a week's allowance in an afternoon without watching it happen. Read how to run Claude Code safely on a VPS before you give an agent a shell on a public box. If you would rather not look after a server at all, Claude Code can also run as a cloud session on infrastructure Anthropic manages; see how cloud sessions are billed.

How Claude Pro usage limits work

Two limits run at the same time, and you can hit either one.

The session limit resets every five hours. Settings > Usage shows the exact time of your next reset, which is the only reliable way to know where you sit in the window. If a banner warns that you've used 90% of your session limit, it is this five-hour window running low rather than the weekly one, so check that reset time before you start anything long.

The weekly limit sits on top of that and applies across all models. Settings > Usage shows when yours resets, and on plans that include Fable it shows a separate weekly reset for Fable as well. As of 9 November 2026 Anthropic's Pro help article says weekly limits reset at a fixed time each week that is assigned to your account, so treat the reset time on your own Usage page as the authority.

Both clocks run on their own, and nothing you do from your side brings a reset forward, so a new chat, a logout or a different browser leaves the counter exactly where it was. The one exception comes from Anthropic: eligible accounts are occasionally given a free limit reset, which appears under Settings > Usage and sets either the session or the weekly limit back to full when you choose to redeem it.

Anthropic publishes no message count for Pro, and that is not evasion. Usage is measured in tokens, so the number of messages you can send varies with message length, including the length of any files you attach. Fifty short questions and five long code reviews are not the same load, which means any figure printed as a fixed message count would be wrong for most people most of the time. People still want a figure anyway, and what a Pro token allowance works out to in practice shows how to read your own consumption instead of chasing a published quota that does not exist. The full mechanics behind Claude's usage limits go deeper than one tier needs.

Usage can also be restricted further at Anthropic's discretion to manage capacity.

What burns through a Pro limit fastest

Long conversations consume more of your usage limit because earlier messages remain part of the context Claude processes. Claude can summarize earlier messages as a conversation approaches its context window, so the full thread does not remain in context indefinitely. Starting a fresh chat for a new topic is the cheapest habit on this page.

After conversation length, these consume the most.

• Attached files. Size counts, and file attachments contribute to the context Claude processes.
• Tools. Web search, Research and MCP (model context protocol) connectors are token heavy, because everything they fetch comes back into the conversation.
• Extended thinking at a high effort level. Higher effort uses more tokens by design, so lower it for routine work.
• Artifacts. Creating and revising them consumes usage.

Two habits claw the usage back. Put reference documents in a project instead of attaching them to every chat, because project documents are cached, and cached portions count less against your limit than new content. And group related questions into one message rather than sending them one at a time, since each round trip carries conversation context again.

Context window against usage limit

These two get confused constantly, and one sentence each sorts it out. The context window controls how long a single conversation can become. The usage limit controls how much you can use Claude across all conversations. With code execution enabled, Claude automatically summarizes earlier messages near the context limit so the chat can continue. Hitting the usage limit pauses new requests until the reset, unless you have usage credits switched on and funded, in which case you keep going at API rates. Claude Code also has a small, capped allowance to finish a step already in progress after hitting the five-hour limit. On Pro it applies up to once per weekly usage period, counts against the weekly limit, and is rolling out gradually.

On paid plans, 200K tokens is now only the fallback for models outside Anthropic's published list, and every current model on that list is larger. In the Claude apps, the newest models (Fable 5.1, Opus 5 and 5.5, Sonnet 5 and 5.5, Haiku 5.5) support a 1M token window on paid plans, while Fable 5 and the 4.x Opus and Sonnet models stop at 500K. In Claude Code the window is larger: every listed model, from Fable 5 and 5.1 through Opus, Sonnet and Haiku 5.5, runs with 1M. The Pro catch is about paying, not size. Sonnet 5 and later and Opus 4.7 and later get the 1M window on Pro with no usage credits needed. Opus 4.6 and Sonnet 4.6 with 1M context still require usage credits on Pro, and so does Fable 5 or 5.1 at any context length, because Fable sits outside Pro's included usage altogether. What one million tokens actually holds puts that number in real terms, and managing context inside Claude Code covers the day to day part.

Usage credits, and paying past the Pro limit

Hitting the limit on Pro leaves you five options: wait for the reset, upgrade to Max or Ultra, turn on usage credits, redeem a free limit reset if Anthropic has given your account one, or stop for the day. If you are staring at the message right now, work out which window you are waiting on before you pick, because the answer to a five hour session cap is rarely the answer to a weekly one.

Usage credits let you keep working past the included allowance. Usage beyond the plan is billed at standard API rates and appears on your bill as a separate line item, apart from the subscription charge. Enable it under Settings > Usage, add a payment method, and set a monthly spending cap. Prepay by adding funds, and switch on auto reload if you would rather not think about it. The phrase does double duty, because the Console sells prepaid API credits under the same name and that balance does nothing for your Pro allowance, so work out which usage credits balance you are funding before you add money. Session limits still reset every five hours as normal. Once you are paying per token, the model you run sets the size of that line, so it is worth knowing what Fable 5 costs per million tokens and which jobs it suits before you hand it a long one. If you would rather stretch the credits on routine work, Haiku 5.5 is the low-cost option, with one catch: any prompt over 100K tokens is billed at five times the normal rate, and a long thread with attachments crosses that line sooner than you would expect.

Set the cap on the first day. Usage credits keep running until you reach your monthly spending cap or your prepaid balance runs out with auto reload off, and the cap can also be set to unlimited. One of those two stops is what stands between an agent stuck in a retry loop and an invoice you did not plan for. Controlling what an AI agent costs you on a VPS covers the larger version of that problem.

Check your real numbers before you upgrade

Settings > Usage shows your current session and weekly consumption along with your next reset time. /status in Claude Code shows the remaining allocation across both surfaces. Watch both for one ordinary working week before you spend more money.

Two patterns mean different things. Running dry every Thursday is a capacity problem, and a bigger plan is the honest fix. Running dry only inside one heavy session is a habit problem, and shorter chats plus project caching fix it at no cost.

You have outgrown Pro when

Four signals, and none of them is a feeling.

• You hit the weekly limit before the week ends, more than once.
• You are paying real money in usage credits every month. Add that overage to $20 and compare the total against the next plan up.
• One task cannot finish inside one session window, so your work spans resets. Large repositories in Claude Code do this first.
• Somebody else needs to use your account. Sharing one login is where the wrong plan starts costing you more than money. Putting two people on one Pro login breaks the consumer terms and leaves both of you drawing on a single usage limit, so price a second seat before you hand over the password.

The first three signals point at Max or Ultra. Max at $100 against Ultra at $200 compares the two plans: as of 9 November 2026, Max costs $100 per month and gives five times Pro's usage every five hours and every week, while Ultra costs $200 per month and gives ten times Pro's usage every week. Those are web prices, mobile pricing may vary, and both plans are billed monthly only. Ultra was formerly called Max 20x, and existing subscribers move automatically with no price change. Max and Ultra both let you spend up to half of their weekly limits on Fable at no extra cost, and they come with $100 and $200 a month in API credits respectively for building on the Claude Platform. Pro gets neither. Max still runs a five-hour session limit and a weekly cap of its own. Verified Ultra accounts have no five-hour limit; accounts not yet verified have a temporary limit. Ultra still has a weekly limit across all models and a separate weekly Fable limit, because neither Max nor Ultra is unlimited, so upgrade for more room rather than expecting the wall to disappear. If you are choosing between vendors rather than between Claude plans at that price, Max and Ultra against ChatGPT Pro at $100 and $200 a month compares how much usage that buys, whether a coding agent is included, and what happens when you hit the ceiling. ChatGPT Pro also now offers a $500 monthly tier. The last signal points at Team, which exists for central billing and admin controls: Claude Team plans for a small business. If the people who need access are a whole department rather than a couple of colleagues, Enterprise swaps the included allowance for per token billing on top of the seat price, which is a different budgeting exercise entirely. Should the decision at that size be between vendors instead, only one of the two publishes an Enterprise seat price at all, which leaves seat minimums, SSO and SCIM, audit logs and retention terms to carry the rest of the comparison. If you do pay for the higher plan and the wall turns up in exactly the same place, check why a new plan's allowance sometimes fails to show up before you conclude the upgrade bought you nothing.

If you are not yet sure Pro is the right place to start, the Claude plan chooser walks the whole ladder instead of one rung of it.

FAQ

How much does Claude Pro cost per month?

$20 per month billed monthly, or $17 per month on annual billing, as listed on Anthropic's pricing page on 9 November 2026. Annual is a single $200 charge for twelve months against $240 paid month by month. Prices vary by region, and the annual plan is charged as a single payment rather than twelve.

Does Claude Pro include Claude Code?

Yes. Claude Code is included from Pro upward, and you log in with the same account you use on claude.ai. The detail that matters is that Claude Code and claude.ai share one usage allowance, so terminal work and browser chats draw down the same limit. Run /status in Claude Code to see what is left before you start a long job.

How many messages do you get on Claude Pro?

Anthropic does not publish a message count, because usage is measured in tokens rather than messages. What you get depends on how long your messages are, including any files you attach, and on whether tools such as web search or Research are running. Its pricing page says Pro gives you at least five times more usage per five-hour session than Free, plus more usage every week. Check Settings > Usage for your own consumption rather than trusting any published estimate.

What happens when you hit the Claude Pro limit?

Unless you have usage credits switched on and funded, Claude stops accepting new messages until the limit resets. The session limit resets every five hours, and the weekly limit resets at the time shown for your account. Both times are shown in Settings > Usage. You can wait for the reset, enable usage credits and continue at standard API rates, redeem a free limit reset if your account has one, or upgrade to Max or Ultra.

Does Claude Pro include API access?

No. Pro covers the Claude apps and Claude Code. API usage is a separate product, set up in the Console and billed per token against its own balance. Paying for Pro gives you no API credit, and API spending does not count against your Pro limits.