Claude Code Pricing: Pro, Max, Team Seats or Pay-Per-Token API

October 15, 2026. Claude Code costs $20 a month on the Pro plan, $100 or $200 a month on Max, $25 or $125 a month per Team seat (less with annual billing), or per token on the API, with no seat fee. Anthropic's own documentation puts the average API cost at about $13 per developer per active day, or $150 to $250 per developer per month across enterprise deployments. Which option is cheapest depends on one number: how many days a month a developer actually codes with it, and how hard. We build AI agents and help companies get recommended inside AI assistants through AI search optimization, so this guide sets out Anthropic's published prices and the arithmetic for choosing between them.

Every price below comes from Anthropic's pricing page and the Claude Code cost documentation, read 15 October 2026. Prices exclude tax and Anthropic says they can change, so check the page before you buy seats for a whole team.

Every way to pay for Claude Code

Claude Code is included in every paid Claude plan. It is not included in the Free plan. The plans differ in how much usage you get, not in whether the tool works.

Two details matter more than the headline prices. Max is billed monthly only, so there is no annual discount. And the Max plans now include monthly API credits for building on the Claude Platform; our guide to what the Max and Team API credits cover explains how far $100 to $500 goes.

How the usage limits actually work

Subscription plans do not bill per token, but they are not unlimited. According to Anthropic's pricing FAQ, every plan has limits that reset on a rolling five-hour window, paid plans add weekly limits on top, and your activity across Claude on the web, desktop, mobile and Claude Code all draws from the same pool. A long coding session therefore eats into the same allowance as your chats.

• Pro gives at least 5x the usage of Free per five-hour session.
• Max gives 5x or 20x the usage of Pro per session, with higher output limits and priority access at busy times.
• Team Standard seats give more than Pro, and Premium seats give 5x more than Standard.
• Fable, which Anthropic calls its most advanced generally available model, is available on Pro only through usage credits; on Max and on Team Premium seats the pricing table lists it at 50 percent of weekly limits.

When you hit a limit you can wait for the reset, upgrade, or on paid plans turn on usage credits and keep working at standard API rates. Anthropic also says it may apply other caps, such as weekly or monthly limits on a model or feature, so there is no fixed number of messages or coding hours per plan.

What Claude Code costs on the API

On the API, Claude Code is billed purely by tokens. These are the current model rates per million tokens:

Three multipliers apply on top. Fast mode for Opus 5.5 costs 2x standard pricing for up to 2.5x the speed. US-only inference costs 1.1x on input and output. And thinking tokens bill as output tokens; the cost documentation notes that thinking cannot be switched off on Opus 5.5, Sonnet 5.5, Haiku 5.5 or the Fable models. Batch processing halves the price, but it suits asynchronous jobs, not an interactive coding session.

Anthropic's average of about $13 per developer per active day is the most useful planning number, and its documentation adds that costs stay below $30 per active day for 90 percent of users. For deeper background on why repeated context is cheap, see our explainer on cached input token pricing.

Subscription or API: the break-even

Divide each plan price by $13 and you get the number of average coding days it pays for at API rates:

• Pro at $20 a month covers about 1.5 average days.
• Max 5x at $100 covers about 7.7 days.
• A Team Premium seat at $125 a month covers about 9.6 days.
• Max 20x at $200 covers about 15.4 days.

A developer who codes with Claude 20 days a month at the average rate would spend about $260 at API prices, more than any single seat. At that rate, Max or a Premium seat costs less on paper. The catch is the usage limit: a seat is cheaper only while your work fits inside its five-hour and weekly windows. Heavy users who keep hitting limits pay the seat price and then usage credits at API rates, so track a few weeks of real usage before you commit a whole team.

The API wins in two cases: occasional users who code with Claude a few days a month, and automated jobs, such as code review in CI, where nobody is sitting at a keyboard and a seat would sit idle.

What makes Claude Code expensive

The cost documentation is unusually candid about where tokens go:

• Long sessions. Claude Code resends the conversation with each request, so a one-line question in a session that has been open all day still pays for the whole history at the cached rate.
• Cache misses. The cache lasts an hour on a subscription but drops to five minutes once you draw on usage credits, and five minutes is the API default. A coffee break can mean reprocessing the full context.
• Agent teams. Teammates running in plan mode use roughly 7x the tokens of a standard session, because each one keeps its own context window.
• Opus as the default. Opus 5.5 costs twice Sonnet 5.5 per token. The cost guide says Sonnet handles most coding tasks well and suggests reserving Opus for complex architectural decisions or multi-step reasoning.
• Context overhead. A large CLAUDE.md loads into every session, and long logs or test output fill the context; the guide suggests hooks that filter output down to errors and skills that load only when needed.

Background housekeeping is small by comparison: Anthropic puts it at typically under $0.04 per session.

How to cap spend for a team

On seat plans, the cost documentation calls the seat allowance the default ceiling: to let developers go past it, an admin turns on usage credits and sets spend limits at the organization, group or member level. Enterprise as now priced on the pricing page adds usage at API rates on top of the seat fee, so check which model your contract uses. The Enterprise consumption guide describes pooled group budgets and alerts at 75 and 90 percent of an organization limit, with limits resetting at 00:00 UTC on the first of the month. On the API, the Claude Console creates a "Claude Code" workspace automatically, and you can put a spend limit on it.

For API teams, Anthropic publishes per-user rate limit starting points that shrink as teams grow, because fewer people code at the same moment:

Which plan to choose

1. Trying it out: Pro at $20 a month, or the API if you only need it a few days a month.
2. One developer, coding daily: Max 5x at $100, moving to Max 20x if the weekly limit bites.
3. A small team that needs central billing and SSO: Team, with Premium seats for the people who code every day and Standard seats for everyone else. Seat types can be mixed.
4. Twenty or more people with security reviews: Enterprise, where you pay $20 a seat plus metered usage and get spend controls, SCIM and audit logs. Our guide to Claude Team and Enterprise pricing compares those plans in detail.
5. Automation and CI jobs: the API, with a workspace spend limit.

If you are a startup, check our AI startup credits comparison before paying list price; credit programs change often. And if your real goal is a working agent rather than a coding tool, our AI engineers build them on these same models. When you want the assistants themselves to recommend your business, that is what our AI search optimization work is for.