Claude Code pricing: plans, API costs and free options

Is Claude Code free?

No. Installing Claude Code costs nothing, but using it does. The official setup docs say Claude Code requires a Pro, Max, Ultra, Team or Enterprise plan, a Claude Console account, or a cloud provider, and that the free claude.ai plan does not include Claude Code access. The pricing page shows the same: Claude Code is marked No for Free and Yes for every paid plan.

That does not mean you need to spend much to find out whether it suits you. There are three cheap or free ways in:

• Learn before you pay. The docs link free self-paced courses, including Claude Code 101, on Claude Academy, and the quickstart and tutorial material is free to read. The Claude Code tutorial on this site covers a first session.
• Try the cheapest paid door. Pro is the smallest plan that includes Claude Code. Annual billing brings it down to USD 17 a month, paid up front as USD 200.
• Pay only for what you use. A Console account bills by token, so a month of light use can cost less than a subscription. Heavy daily use usually does not (see the API section below).

If your goal is a coding agent at no cost at all, Claude Code is the wrong tool. The Claude Code alternatives page covers other options and their trade-offs.

Which Claude plans include Claude Code?

Every paid plan includes Claude Code. It shares the same usage pool as your chats on web, desktop and mobile, so a long coding session and a long conversation draw from one allowance. Prices below are from the Claude pricing page as of October 2026 and do not include tax; Anthropic can change them.

Team plans are listed for teams of 2 to 150 people, and you can mix seat types. The Max and Ultra plans also list a monthly allowance of Claude Platform API credits, USD 100 and USD 200. Read the next section before counting those toward your Claude Code costs.

Choosing between the two individual tiers is its own question, covered in Claude Pro vs Max. Whether the spend pays off is a separate question again; see whether Claude Code is worth it.

What are the usage limits on a Claude plan?

Claude Code does not bill you per message on a subscription. It draws from your plan's allowance, and that allowance resets on two clocks. Every plan has a rolling five-hour session window, and paid plans add weekly limits on top. The pricing page states there is no fixed message count, because how fast you reach a limit depends on the model you use, how long your conversations are and which features you use.

• Where you stand: type /usage inside Claude Code. On Pro, Max and Ultra it shows plan usage bars, activity stats and a breakdown of what counts against your limits. You can also check Settings, then Usage, on claude.ai.
• When you hit a limit: wait for the reset, move to a higher plan, or turn on usage credits to keep working at standard API rates. Run /usage-credits inside Claude Code to manage them; it opens the usage settings in your browser. It is not available when you authenticate with an API key.
• Teams: each member's Claude Code usage draws from a per-seat allowance that resets on a rolling five-hour window and a weekly window, shared with Claude chat and Cowork.

A longer explanation of limits, and how to stretch them, is in the guide to Claude Code usage limits.

How does pay-as-you-go API pricing work for Claude Code?

If you log in with a Claude Console account instead of a subscription, Claude Code bills by token at API list prices. The first time you authenticate with Console, a workspace called Claude Code is created automatically, which gives you cost tracking and a place to set spend limits. API access in the Console runs on pre-paid credits.

Prices are per million tokens (MTok). These are the standard rates from the Claude Platform pricing page as of October 2026, in US dollars:

Those numbers alone are hard to turn into a monthly bill, so here is arithmetic, not a forecast. Suppose a work session sends one million input tokens and receives 100,000 output tokens, with no caching. At list price that costs USD 3.00 on Sonnet 5.5 (USD 2.00 for input plus USD 1.00 for output), USD 6.00 on Opus 5.5 and USD 15.00 on Fable 5.1. In practice Claude Code caches repeated context, and cached reads cost a fraction of the input price, so real sessions with the same token volume cost less. Thinking tokens bill as output tokens.

For a real-world anchor, the Claude Code cost documentation reports that across enterprise deployments the average is around USD 13 per developer per active day and USD 150 to USD 250 per developer per month, with costs below USD 30 per active day for 90 percent of users. Treat that as a team-level average, not a promise for you. Your number depends on the model, the codebase and how often you run parallel sessions or automation.

Inside a session, /usage shows a token breakdown and a dollar estimate computed locally from list prices. For billing you can trust, use the Usage page in the Claude Console. If you want to use your own key, the Claude Code API key guide covers setup.

Which option should you choose?

Match the option to how you work, not to the biggest number on the page.

A simple rule: start on the smallest plan that covers your week, watch /usage for two weeks, and move up only if you actually run into the limit. Switching plans is easy. Paying for headroom you never use is the common waste.

How do you make Claude Code cost less?

Whether you pay per token or draw on a plan, usage scales with how much context Claude processes. These habits come from the official cost-management page and apply to both.

• Clear between tasks. Run /clear when you switch to unrelated work. Stale context costs tokens on every later message.
• Pick the right model. The docs say Sonnet handles most coding tasks well and costs less than Opus; reserve Opus for hard architectural decisions. Switch with /model.
• Plan before you build. Press Shift+Tab to cycle to plan mode for complex work, so Claude proposes an approach before editing and you avoid expensive rework.
• Write specific prompts. The docs contrast "improve this codebase", which triggers broad scanning, with "add input validation to the login function in auth.ts".
• Stop early. If Claude heads the wrong way, press Esc, then use /rewind to restore the conversation and code to an earlier checkpoint.
• Keep CLAUDE.md lean. It loads at the start of every session. Move long, task-specific procedures into skills, which load only when used. See the guide to writing CLAUDE.md.
• Lower the effort for simple tasks. Extended thinking is on by default and bills as output tokens; /effort lowers it.

Agent teams are the one feature that can multiply a bill quickly. The docs say they use roughly 7x more tokens than a standard session when teammates run in plan mode, because each teammate keeps its own context window.

What do you do if you want to build something and spend little?

Decide what to build before you subscribe. A month of Pro pays off only if you arrive with a project, and the most common reason people waste a subscription is not having one. Pick a small idea you can finish in a weekend and scope it so the first version is one workflow.

The board is a free place to find one. Browse all ideas or a category such as developer tools, pick something other people posted and will not build, and copy its generated build prompt into Claude Code. If you have an idea you will never build yourself, you can post it at no cost. For the build itself, see how to build an app with Claude Code, and for installation, how to install Claude Code.