Claude Code Subscription: Plans and Prices (2026)

What is a Claude Code subscription?

There is no standalone Claude Code subscription. Claude Code is included in whatever Claude plan you already pay for, so you subscribe to Claude (Pro, Max, Team, or Enterprise) and Claude Code comes with it at no extra cost. The same monthly fee covers the claude.ai chat interface, the desktop and mobile apps, and the claude terminal command, all drawing from one shared usage pool. The entry point is Claude Pro at $20/month; heavier users move to Max at $100 or $200/month. There is also a separate pay-per-token API path for automation. This guide shows every tier, the real usage limits, and how to pick and activate the right one in 2026.

Claude Code subscription plans at a glance

Every tier below includes Claude Code. The price you pay is for the Claude plan; the terminal agent is bundled in. Annual billing lowers the monthly rate on most plans.

Plan	Price (monthly)	Price (annual)	Claude Code included?	Best for
Free	$0	$0	Yes, limited usage	Trying Claude Code on small tasks
Pro	$20/mo	$17/mo	Yes	Solo developers, a few hours a day
Max 5x	$100/mo	not discounted	Yes, plus Opus access	Full-time daily use
Max 20x	$200/mo	not discounted	Yes, plus Opus access	All-day and parallel sessions
Team Standard	$25/seat/mo	$20/seat/mo	No (chat only)	Non-engineering teams
Team Premium	$125/seat/mo	$100/seat/mo	Yes	Engineering teams of 5+
Enterprise	Custom	Custom	Yes	50+ developers, compliance needs
API (pay per token)	Usage-based	Usage-based	Yes, for automation	Scripts, CI/CD, products

Two things to flag before you read on. Team Standard is the one tier that does not include Claude Code (it is chat access only), so engineering teams want Team Premium. And the API row is not a subscription at all, it is metered usage, covered in its own section below.

How the Claude Code subscription actually works

Anthropic runs two completely separate billing systems, and Claude Code can be paid for through either.

The subscription path (most developers). You pay a flat monthly fee for a Claude plan, and that fee unlocks Claude everywhere: the web app, the desktop and mobile apps, and the claude command in your terminal, all drawing from a single shared usage pool that refills on a rolling window. This is what people mean by "Claude Code subscription," even though they are really subscribing to Claude.

The API path (automation only). You pay per token with no monthly fee. It is required for scripted, automated, or CI/CD use, because the subscription terms prohibit unattended automation. The two systems do not overlap: a Max subscription grants zero API tokens, and an API balance grants zero subscription features.

One trap is worth knowing now. If the ANTHROPIC_API_KEY environment variable is set in your shell, Claude Code bills at per-token API rates and ignores your subscription entirely, so developers with an old key in their dotfiles can burn API credits without realising it. To use your subscription, make sure that variable is unset. For both systems side by side, see our Claude Code pricing guide.

Free: can you run Claude Code without paying?

Yes, with real limits. A free Claude account can authenticate Claude Code through OAuth and run it on small tasks: enough to see how the terminal agent behaves, read a file, make a contained edit, or ask questions about a small codebase.

What free will not do is carry sustained work. The usage budget is small, you do not get the Opus model, and you hit the ceiling quickly on anything beyond a quick experiment. Treat it as an evaluation, not a workflow. If you would rather evaluate via the API instead, new Anthropic API accounts get $5 in starter credits (after phone verification), enough for a handful of sessions.

Pro: the $20 entry point

Pro at $20/month (or $17/month billed annually) is where most individual developers land. It includes Claude Code in the terminal alongside the web, desktop, and mobile apps, all sharing one usage pool.

What you get:

• Claude Code terminal access with Sonnet 4.6 as the default model
• A meaningfully larger usage budget than free, on a rolling 5-hour window
• Extended thinking mode and unlimited projects
• Priority over free-tier users during high demand

The usage model. Pro runs on a rolling 5-hour window plus a 7-day weekly ceiling. When you exhaust the budget, you wait for the window to roll forward (typically 30 to 60 minutes) rather than hitting a hard wall for the day. Sustained heavy use will reach both limits.

Who it fits. Developers using Claude Code for a few focused hours a day, smaller projects, and anyone testing the tool before committing to more. The rule of thumb: if you are not regularly hitting rate limits, Pro is correct. If you are, Max is the fix. Our Claude Code Pro plan guide covers exactly where the Pro ceiling sits and the signals that you have outgrown it.

Max: $100 and $200 for heavy users

Max is the subscription tier for developers who treat Claude Code as a primary tool. It multiplies the Pro usage budget and adds the Opus model plus priority routing.

Max 5x ($100/month): five times the Pro usage per session, full access to the Opus model (the most capable for architecture and complex refactors), priority traffic routing during peak demand, and everything in Pro. This is the right tier for full-time developers who hit Pro limits daily.

Max 20x ($200/month): twenty times the Pro usage, plus everything in Max 5x. This is for developers running Claude Code all day, every day, or running multiple concurrent sessions where a mid-task rate limit is genuinely costly.

Is Max worth it? It comes down to one question: how often do you hit Pro limits? Rarely, and Pro is right. Daily, and Max 5x pays for itself fast once you price in the lost flow state from a blocked tool, with more Opus usage as a real bonus on top. For developers who hit Pro limits daily, Max 5x at $100/month usually costs less than the equivalent API tokens would, which is what makes the flat subscription worth it for heavy interactive use.

Team and Enterprise: subscriptions for organizations

Team plans add centralized billing, admin controls, and SSO on top of individual features. There are two seat types, and the distinction matters.

Team Standard ($25/seat/month, $20 annually): chat access only, roughly 1.25x Pro usage per seat, centralized billing, admin dashboard, SSO, 5-seat minimum. This tier does not include Claude Code, so it suits non-engineering teams (marketing, product, support) that need Claude but not the terminal agent.

Team Premium ($125/seat/month, $100 annually): full Claude Code access, roughly 6.25x Pro usage per seat, all the management features, 5-seat minimum. This is the tier for engineering teams. At $125/seat it costs more than an individual Max 5x ($100), but you get the admin controls and usage visibility that individual subscriptions lack.

Enterprise (custom pricing): for 50+ developers or organizations with specific security and compliance needs. It adds a 1M-token context window (versus 200K on consumer plans), SCIM provisioning, audit logging, compliance APIs, and dedicated support. Reported 2026 figures range from a few hundred dollars a month for small deployments up to five figures for large engineering organizations.

One rule applies to every paid tier: seat sharing is prohibited. Anthropic's terms ban credential sharing, and flagged accounts get terminated.

The API path: not a subscription

The API is the alternative to a subscription, not a tier of one. You pay per token, with no monthly fee, and you use it whenever Claude Code runs unattended: scripts, hooks that call Claude, CI/CD pipelines, or a product you are building on top of the model.

Current API rates (2026):

Model	Input (per 1M tokens)	Output (per 1M tokens)
Claude Opus 4.8	$15	$75
Claude Sonnet 4.6	$3	$15
Claude Haiku 4.5	$1	$5

New API accounts get $5 in free credits after phone verification, then rate limits scale with cumulative spend. The hard rule, again: subscriptions and API access are entirely separate. If you automate Claude Code in any way, you need a funded API key, not a subscription, and you manage the two billing systems independently.

Which Claude Code subscription should you choose?

Match the plan to how hard you actually use the tool, not how hard you imagine you will.

Your situation	Right subscription
Just trying Claude Code on small tasks	Free
Solo developer, a few hours a day, rarely rate-limited	Pro ($20/mo)
Daily use, hitting Pro limits regularly	Max 5x ($100/mo)
Claude Code is your main tool, full-day or parallel sessions	Max 20x ($200/mo)
Engineering team of 5+, needs central billing and admin	Team Premium ($125/seat/mo)
50+ developers or compliance requirements	Enterprise (custom)
Automation, scripts, CI/CD, product integration	API (pay per token)

The honest default is to start one tier lower than you think you need. Run Pro for a full week of normal work. If you hit rate limits more than a couple of times, upgrade to Max 5x. Almost nobody needs Max 20x unless they are running Claude Code all day or several sessions at once. Buying headroom you never touch is the most common way developers overpay. For the full decision tree across solo, team, and enterprise, see our Claude Code plans guide.

How to activate your Claude Code subscription

Because the subscription rides on your Claude account, activation is two steps, not a separate checkout.

1. Subscribe to a Claude plan. Go to claude.ai, sign in, and pick Pro, Max, or your team's plan. This is the only place you pay; there is no separate Claude Code purchase.
2. Install Claude Code. It runs from your terminal and is installed with npm:

   npm install -g @anthropic-ai/claude-code
   

   It requires Node.js 18 or later (check with node --version). For Windows, Linux, and the common install errors, our install Claude Code guide has the platform-specific steps.
3. Authenticate with your subscription. Run claude in any project directory. On first launch it opens an OAuth login in your browser. Sign in with the same account you subscribed with, and the terminal agent now draws from your subscription's usage pool.
4. Confirm you are not on API billing. If you want subscription billing, make sure ANTHROPIC_API_KEY is not set in your shell, since its presence silently switches Claude Code to per-token API charges.

That is the whole path: subscribe once, install once, authenticate once. If you are new to the tool, what is Claude Code explains what the agent actually does before you pick a tier.

Where the real cost hides: configuration

The subscription is the easy part of the cost. What most developers underestimate is configuration: a bare Claude Code install is capable but generic. It does not know your codebase conventions, remember last week's decisions, or follow your deploy process. Closing those gaps with CLAUDE.md files, persistent memory, custom skills, and hooks is where the time goes, and it is what actually determines output quality.

This matters for your tier choice because configuration changes the return on every token. A well-configured Pro subscription routinely outperforms a default Max subscription, since output quality is set by configuration more than raw token limits. Tuning your setup is often a better investment than upgrading.

That configuration layer is what Claudify provides as a one-time purchase: a pre-built operating system for Claude Code (skills, persistent memory, hooks, and specialist agents) that installs in one command. It does not replace your Claude subscription, it raises the value of whichever tier you are on, and it is a one-time fee rather than another subscription.

Frequently asked questions

Is there a separate subscription just for Claude Code?

No. Claude Code is not sold on its own. It is bundled into your Claude subscription, so when you pay for Claude Pro, Max, Team Premium, or Enterprise, the terminal agent is included at no extra charge and shares the same usage budget as the chat interface. Team Standard is the one paid tier that excludes Claude Code. The only other way to access Claude Code is the pay-per-token API, which is metered usage rather than a subscription.

How much does a Claude Code subscription cost?

The cheapest paid entry is Claude Pro at $20/month (or $17/month billed annually), which includes Claude Code. Heavier users move to Max 5x at $100/month or Max 20x at $200/month, which multiply the usage budget and add the Opus model. Teams pay $125/seat/month for Team Premium (the seat type that includes Claude Code), and Enterprise is custom-priced. A free Claude account can also run Claude Code on small tasks at no cost.

Can I use one Claude Code subscription across my whole team?

No. Each developer needs their own subscription or team seat. Anthropic's terms explicitly prohibit credential sharing, and accounts flagged for multi-user access are terminated. For teams of five or more, Team Premium ($125/seat/month) adds centralized billing, admin controls, and usage visibility on top of individual access. For very small teams, individual Max subscriptions can be simpler.

Does a Claude Code subscription include API access for automation?

No, and this catches people out. A subscription covers interactive use in the terminal and apps only. Automated, scripted, or CI/CD use requires a separate pay-per-token API key, because the subscription terms prohibit unattended automation. If ANTHROPIC_API_KEY is set in your shell, Claude Code bills at API rates and ignores your subscription, so keep the two billing paths deliberately separate.

Is a Claude Code subscription worth it versus tools like Cursor?

For terminal-first, autonomous work, yes. A $20 Pro subscription competes with Cursor Pro on price but operates differently: Claude Code is an agent that reads, writes, debugs, and runs multi-step tasks from the command line, where Cursor is an IDE plugin enhancing your editor. The right choice depends on workflow, not just price. Our Claude Code vs Cursor comparison breaks down where each one wins.

Summary

The phrase "Claude Code subscription" is slightly misleading, and clearing that up is the point of this guide: you do not buy Claude Code on its own. You subscribe to Claude, and Claude Code is included.

• Free: limited Claude Code on small tasks, good for evaluation only.
• Pro ($20/mo, $17 annually): the real entry point, fine for a few hours of daily use.
• Max 5x ($100/mo): for developers hitting Pro limits daily, adds Opus access.
• Max 20x ($200/mo): for all-day or parallel use where rate limits are costly.
• Team Premium ($125/seat/mo): the team tier that includes Claude Code.
• Enterprise (custom): 50+ developers, compliance, 1M context window.
• API (pay per token): the non-subscription path, required for automation.

Pick the tier that removes rate-limit interruptions without paying for headroom you never use, and start one tier lower than you expect. Then remember that the bigger lever, at any price point, is configuration. A well-tuned setup changes what every token returns. If you want that setup ready in one command, get Claudify and point it at whichever Claude plan you are on.