All articles

How Much Does Claude Code Cost per Month? $20, $100, $200 Plans vs API (Nov 2026)

Claude Code has no price of its own: it comes with Pro ($20), Max ($100) and Ultra ($200), or you pay per token through the API. We checked every price against Anthropic's pages on Nov 10, 2026 and tested that the cost Claude Code displays follows the official token prices.

Claude Code cost per month: a pale blue clay terminal block next to a stack of beige clay coins and a small price tag, soft light and leaf shadows
Also available in中文Español

The Claude Code cost per month isn’t a separate line item. If you use it yourself, you pay for a Claude subscription: Pro is $20 a month ($200 billed annually, about $17 a month), Max is $100 a month and Ultra is $200 a month. All three include Claude Code and share one usage pool with chat on the web. On Team, a Standard seat is $20 a month billed annually ($25 monthly) and a Premium seat is $100 billed annually ($125 monthly). If you pay per token through the API instead, Anthropic’s own average for enterprise deployments is about $13 per developer per active day and $150 to $250 per developer per month. Every price here was checked against Anthropic’s official pages on November 10, 2026, in USD before tax. We also tested one thing on our own machine: the cost Claude Code displays is computed from Anthropic’s official token prices.

The whole picture: personal, team, enterprise and API

Prices come from claude.com/pricing; whether Claude Code is included comes from the plan comparison table on the same page.

Claude Code price per month by plan on Nov 10, 2026: Free no Claude Code, Pro $20, Max $100, Ultra $200, Team Standard $20 per seat, Team Premium $100 per seat, Enterprise $20 per seat plus usage, API per token

A few things that are easy to misread:

  • The Free plan has no Claude Code. In the official comparison table the Claude Code row says Yes only for Pro, Max and Ultra.
  • Anthropic says Claude Code is included in all paid plans and shares the same limits as your chats on the web, desktop and mobile. For heavy coding you can also switch to pay-as-you-go API credits through a Console account.
  • The help center articles for Max and Ultra say the published price is for web subscriptions only and mobile pricing may differ, so trust what the app shows.
  • Max and Ultra include $100 and $200 a month in API credits, but those can’t be used in interactive Claude Code and can’t pay for usage credits. See Claude Pro vs Max vs Ultra.

Pay per token: price per million tokens

On Pro, Max and Ultra the default model in Claude Code is Opus 5.5 (see the model configuration docs). Anthropic’s API pricing page lists the models you are most likely to use, in USD per million tokens:

ModelInputOutputCache hits (read)
Opus 5.54200.20
Sonnet 5.52100.10
Fable 5.28400.20
Haiku 5.5 (prompts up to 100K tokens)0.100.500.01

For Haiku 5.5, prompts over 100K tokens move to $0.50 input and $2.50 output. The pricing page also says fast mode for Opus 5.5 and Fable 5.2 is billed at 2x the standard price.

Anthropic’s average, and our conversion

The official cost management docs say that across enterprise deployments the average is around $13 per developer per active day and $150 to $250 per developer per month, with costs below $30 per active day for 90% of users. They add that costs vary widely with model, codebase size and whether you run several instances or automation, and suggest a small pilot first.

Put side by side: $13 times 20 active days is $260, a little above the documented monthly range, and $150 to $250 a month corresponds to roughly 12 to 19 active days a month. That division is ours, not Anthropic’s.

Here is another conversion from the official token prices. Take a fairly long Opus 5.5 session with 3 million input tokens and 150,000 output tokens, no caching: 4 × 3 + 20 × 0.15 = $15. The same session on Sonnet 5.5 is $7.50. That’s our assumption, not an official figure, and real bills depend heavily on caching and context length.

Estimated monthly API cost of long Claude Code sessions on Opus 5.5 at $15 each and Sonnet 5.5 at $7.50 each, compared with Pro $20, Max $100 and Ultra $200

Under that assumption, seven such Opus 5.5 sessions a month already pass Max’s $100 (7 × $15 = $105); on Sonnet 5.5 it takes 14. A few scattered sessions are cheaper on the API; heavy daily use is cheaper on a subscription.

What to pay per month, by how you work

  • A few times a week, or mostly your own scripts: pay per token through the API and set a spend limit in the Console.
  • Daily coding, hitting the limit less than twice a week: Pro, $20 a month.
  • Hours of Claude Code on most workdays: Max, $100 a month.
  • Several agents in parallel, or lots of Fable: Ultra, $200 a month.
  • A team: Team Standard seats from $20 per seat per month billed annually, Premium seats for more usage. If you need SSO or audit logs, Enterprise is $20 per seat per month plus usage at API rates, billed annually with a 20-seat minimum.

If you hit the wall, see the Claude Code session limit message. For subscriptions and API prices side by side, see AI subscription and API price comparison. Against Codex, see Claude Code vs Codex.

What happens past your plan’s allowance

With usage credits on, hitting the limit moves your requests to pay-as-you-go at standard API rates instead of blocking them. You set the monthly spend limit yourself, and credit purchases are non-refundable. The details are in the session limit article above.

If an ANTHROPIC_API_KEY environment variable is set, Claude Code uses that key and bills API usage instead of your subscription allowance. Anthropic’s help center calls this out on the Claude Code plan page. If you’re a subscriber trying to save money, check for a leftover variable first.

What we tested and what is from the docs

We tested one thing: whether the cost Claude Code shows follows the official prices. We ran a fake API server on the box, had Claude Code 2.1.322 send requests with a fake API key, made the server report a fixed usage (for example 1 million input and 100,000 output tokens), and read total_cost_usd from claude -p --output-format json.

Claude Code 2.1.322 total_cost_usd against a fake API server: Opus 5.5, Sonnet 5.5, Fable 5.2 and Haiku 5.5 costs match the official price table

The results match the official price table for Opus 5.5, Sonnet 5.5, Fable 5.2 and Haiku 5.5, and cache reads are charged at the cache-read price. Our Haiku 5.5 run used 1 million input tokens, so it landed in the “prompts over 100K tokens” tier and was priced at $0.50 and $2.50.

That figure is Claude Code’s local estimate at list price. Anthropic’s docs say the cost shown to subscribers isn’t a billing amount, and organizations on contracted rates will see different numbers. We produced no real bill.

FAQ

Is Claude Code free?

Not on the Free plan. You need at least Pro ($20 a month), or a Console account paying per token.

Why doesn’t Claude Code have its own price on a subscription?

Anthropic counts it inside the plan allowance: on Pro, Max and Ultra, Claude Code and chat share one 5-hour and one weekly allowance, so there is no separate Claude Code price.

Is $20 a month enough?

It depends on how often you hit the limit. Anthropic publishes no message counts, only that usage depends on conversation length, model, attachments and features. Our rule of thumb is in Claude Pro vs Max vs Ultra.

Can I treat Total cost in /usage as a bill?

Not as a subscriber. Anthropic says subscribers’ usage is included in the subscription, so the session cost figure isn’t for billing; it is estimated locally at list price.

Do Team seats get a different Claude Code allowance?

Each Team seat has its own allowance that resets on the same 5-hour and weekly windows, sized by seat tier: Standard gives more than Pro and Premium gives 5x Standard.

Leave a thought

Your email address will not be published.