Claude Code Pricing and Limits (2026)

Claude Code starts at $20 per month for an individual Pro plan. Max 5x costs $100 per month and Max 20x costs $200 per month. Those subscriptions include interactive Claude Code usage within rolling and weekly limits; API-key users pay separately by token.

The confusing part is not the sticker price. Claude and Claude Code share one usage pool, Anthropic does not promise a fixed message count, and usage credits can turn a fixed subscription into variable spend after you hit a limit. This guide separates those systems and shows how to calculate a pay-as-you-go coding session from actual billed tokens.

Claude Code Pricing: Short Answer

Most individuals should start with Pro at $20 per month, then upgrade only if rolling or weekly limits regularly interrupt paid work. Max 5x and Max 20x buy more capacity, not unlimited usage.

• Free: $0, but Claude Code is not included.
• Pro: $20 monthly or $200 annually.
• Max 5x: $100 monthly with five times Pro capacity per session window.
• Max 20x: $200 monthly with twenty times Pro capacity per session window.
• API or usage credits: variable token billing at the selected model's rates.

These are Anthropic's current US web prices as of July 30, 2026. Taxes, mobile-app pricing, regional availability, and plan terms can differ. Check the official Claude pricing page before purchasing.

Claude Code Plan Comparison

The individual plans differ mainly in included capacity and billing interval. Every paid individual plan includes Claude Code and shares usage across Claude web, desktop, mobile, and terminal sessions.

Is Claude Code Free?

No. The Claude Code installer is free to download, but the free Claude.ai plan does not include Claude Code access. You need a paid Claude plan, a Claude Console account with API billing, or a supported cloud provider account.

Installation does not start a subscription by itself. Follow the Claude Code install guide to set up the binary, then choose subscription login or API billing during authentication.

How Claude Code Usage Limits Work

Subscription limits reset on a rolling five-hour session window, and paid plans also have weekly limits. Claude does not publish a fixed number of messages because usage depends on model choice, repository size, context length, conversation complexity, and feature use.

Activity across Claude web, desktop, mobile, and Claude Code draws from the same subscription pool. A long research conversation can therefore reduce the capacity available for a terminal session later in the same window.

Max 5x provides five times Pro capacity per five-hour window; Max 20x provides twenty times Pro capacity. Those labels do not remove weekly caps, model-specific restrictions, or other controls Anthropic may use to manage capacity.

Anthropic explains the shared pool and reset behavior in its Pro and Max Claude Code guide and the Claude Code models and limits reference.

Which Claude Code Plan Should You Choose?

Choose a plan from measured interruptions, not from a guessed message count. Start with the lowest tier that supports Claude Code, work for a normal billing cycle, and note how often limits block valuable work.

Claude Pro

Pro is the right starting point for individuals who use Claude Code regularly but do not run it continuously all day. It costs $20 monthly or $200 annually and includes the rest of Claude in the same plan.

The tradeoff is capacity. Large repositories, long conversations, and repeated tool calls consume the shared pool faster than short, well-scoped tasks. If Pro works most days, occasional waiting is cheaper than paying five times as much for Max.

Claude Max 5x

Max 5x is for frequent users who repeatedly exhaust Pro during paid work. It costs $100 per month and increases per-session capacity to five times Pro.

The upgrade costs $80 more than monthly Pro. Before paying it, compare that amount with the value of interrupted time and the actual usage credits or API spend needed to finish the work.

Claude Max 20x

Max 20x is for daily users whose workflow consumes far more than Max 5x. It costs $200 per month and increases per-session capacity to twenty times Pro.

It is still limited. Teams, automated systems, and workloads that need predictable throughput should evaluate API or organization plans instead of treating Max 20x as unlimited infrastructure.

Team and Enterprise

Team plans add central billing, administration, and organization controls for groups, while Enterprise pricing is sales-led. Claude Code is included in Team seats, but seat prices and included capacity differ from individual Pro and Max.

Anthropic currently lists Team Standard at $25 per seat monthly or $20 per seat monthly when billed annually, and Team Premium at $125 monthly or $100 when billed annually. Verify current seat minimums, controls, and enterprise terms on the official pricing page.

Claude Code API Pricing

API-key usage is billed by aggregate input and output tokens rather than a fixed monthly allowance. The selected model, prompt-cache use, extended thinking, tool definitions, and every conversation turn affect the bill.

Sonnet 5's $2 input and $10 output rates are introductory through August 31, 2026; Anthropic lists standard pricing of $3 input and $15 output per million tokens after that date. Check the official Claude API pricing table for current model rates and caching multipliers before estimating a long-running job.

Tool use adds its schemas and results to token usage. Extended thinking tokens are billed as output even when the full reasoning is not shown. Prompt-cache reads cost ten percent of base input on the models above, which can materially reduce repeated repository context.

What Does a Coding Session Really Cost?

A pay-as-you-go coding session costs the sum of all billed input, cache, and output tokens across its turns. A 200,000-token context window does not mean one 200,000-token charge; the bill uses the aggregate tokens actually processed in every request.

cost =
  uncached input tokens / 1,000,000 * input rate
  + cache read tokens / 1,000,000 * cache rate
  + output tokens / 1,000,000 * output rate

These examples use aggregate tokens and exclude cache writes, server-side tool charges, cloud-provider markups, and taxes. They are calculations, not promises about how many tokens your repository will consume.

A subscription does not assign a dollar price to each interactive session. You pay the fixed plan amount until you hit included limits, then either stop, upgrade, or deliberately enable variable usage credits.

What Happens When You Hit a Limit?

When a subscription limit is reached, you can wait for it to reset, upgrade, or enable usage credits and continue at standard API rates. Claude shows remaining capacity warnings before the hard stop.

Usage credits are separate from the subscription price. If enabled, they prevent an interruption by creating metered overage spend. Set a monthly cap and alerts before turning them on, especially if an agent can run unattended.

Anthropic also sells prepaid usage bundles with volume discounts. Bundle terms and purchase caps can change, so compare the current effective price with upgrading a subscription.

The Cheapest Way to Use Claude Code

Annual Pro at $200 per year is the cheapest effective individual subscription with Claude Code included, about $16.67 per month. Monthly Pro costs $20. Very light API usage can cost less than either option, but its bill is variable and you must supply a Console or cloud-provider account.

The cheapest durable strategy is to start on Pro, keep tasks scoped, start fresh sessions when context is stale, and upgrade only after measuring real limit interruptions. API users should prefer the least expensive model that completes the task and preserve prompt-cache hits on repeated context.

Model choice matters as much as plan choice. The current AI model API ID reference lists the exact Claude aliases and snapshots to verify before setting a production model.

Is Claude Code Worth Paying For?

Claude Code is worth paying for when repository-aware edits, command execution, and agent workflows save more than the plan costs. Pro is a low-risk test; Max only pays off when capacity, not product fit, is the recurring problem.

Compare actual workflows rather than chatbot benchmarks. My Claude Code vs Codex vs Cursor comparison explains where each product fits, and the complete Claude Code guide shows the features you are paying to use.

Integrations can also change the value. An MCP server can add external tools and data to the same session, but each tool schema and result also consumes context. More capability is useful only when it helps the current task.

How to Track Usage and Cost

Run /usage inside Claude Code to inspect plan usage or API-token estimates, then use the Claude Console Usage page as the authoritative billing record for metered accounts.

/usage
/model
/context

/model shows which model is active, while /context reveals what is consuming the context window, including MCP tool definitions. If you need to audit a completed session, the client-side Claude Code Session Viewer parses the JSONL transcript, tool calls, tokens, and estimated cost without uploading the file.

Anthropic's Claude Code cost management guide documents the current commands, caching guidance, spend controls, and the difference between subscription usage and API billing.