Belum ada versi bahasa Melayu, jadi dipaparkan dalam bahasa Inggeris.

Claude Code pricing (November 2026): every plan and cost

Convika Team19 min read

  • claude-code
  • ai-agents
  • ci
On this page

Claude Code costs $20 a month on Pro, $100 on Max, and $200 on Ultra, and the Free plan doesn't include it. With an API key you pay per token instead: $4 in and $20 out per million tokens on Opus 5.5, the default model. Anthropic puts the API average at $150 to $250 per developer per month.

Every Anthropic and GitHub number below was read on its own page on November 18, 2026, between 8:23 and 8:27 p.m. Japan time, and each page is linked under Sources.

  • Plans: Claude Pro is $20 ($200 a year), Max $100, Ultra $200. Team is $25 a seat, or $125 for a premium seat. Enterprise is $20 a seat plus usage at API rates.
  • Limits: a 5-hour limit and a weekly limit, shared by Claude chat and Claude Code. Ultra has no 5-hour limit on verified accounts.
  • Past the limit: wait, upgrade, or turn on usage credits and keep working at API rates under a monthly cap you set.
  • The other bill: an agent that opens pull requests starts GitHub Actions runs, and GitHub bills those minutes. Skipbox runs them for a flat $39 a month with unlimited minutes.

Claude Code pricing plans

Anthropic's pricing page, as a table. Prices are in US dollars before tax.

PlanMonthlyAnnualUsageClaude Code
Free$0NoneLimitedNot included
Pro$20$200 a year, about $17 a monthMore than FreeIncluded
Max$100None5x Pro, every 5 hours and every weekIncluded
Ultra$200None10x Pro every week, no 5-hour limit on verified accountsIncluded
Team, standard seat$25 a seat$20 a seat a monthMore than ProIncluded
Team, premium seat$125 a seat$100 a seat a month5x a standard seatIncluded
EnterpriseAnnual only$20 a seat a month, 20-seat minimum, plus usage at API ratesNo per-seat limitIncluded
API key (Claude Console)Per tokenNoneNo plan limitIncluded

Claude Code also runs through Amazon Bedrock, Google Cloud's Agent Platform, and Microsoft Foundry. Those bill tokens to your cloud account.

Is Claude Code free?

No. Anthropic's setup page says: "The free claude.ai plan does not include Claude Code access." You need a Pro, Max, Ultra, Team, or Enterprise plan, or a Console account that pays per token.

The cheapest way in is Pro at $20 a month. Paid yearly it is $200 up front, about $17 a month.

Is Claude's $20 Pro plan enough?

Pro runs the same Claude Code as every other plan, on the same default model, Opus 5.5. What changes is how long you work before a limit stops you.

Anthropic gives no message count: "there's no fixed message count," because usage depends on "the length and complexity of your conversations, the model you choose, and the features you use." Two things make Pro run out sooner in Claude Code than in chat:

  • One pool. "Your activity across Claude on web, desktop, mobile, and Claude Code all draws from the same pool." A morning of chat comes out of the same 5 hours as your terminal.
  • Heavy turns. Anthropic's cost guide: "each Claude Code turn carries file contents, tool calls, and multi-step reasoning, so one debugging session can consume more than a day of chat."

Anthropic's own advice to Pro users is to move to Max "if you consistently hit limits and need more capacity for larger repositories." Sessions in parallel multiply the draw, so agents that run side by side for hours are what Max and Ultra are sized for. Fable, the top model, is paid from usage credits on Pro.

Pro vs Max vs Ultra

ProMaxUltra
Price$20 a month$100 a month$200 a month
5-hour limit1x5xNone on verified accounts
Weekly limit1x5x10x
FablePaid with usage creditsIncluded, up to half your weekly limitIncluded, up to half your weekly limit
Monthly API creditsNone$100$200
Annual billingYesNoNo

Max multiplies both Pro limits by 5. Ultra doubles Max's week and removes the 5-hour limit, so a long agent run doesn't stop at hour 5. Anthropic verifies Ultra accounts automatically from account and payment history, and a temporary limit applies until it does.

What happened to Max 20x?

It is Ultra now. Anthropic's help center: "The Max 20x plan is now called the Ultra plan. Existing Max 20x subscribers move to Ultra automatically, with no price change." The price stays $200 a month. What you get is stated differently: "10 times the Pro plan's usage every week" and no 5-hour limit on verified accounts. The plan once called Max 5x is today's Max.

Do the monthly API credits work in Claude Code?

No. Max, Ultra, and Team include monthly credits for the Claude Platform, and the help center is direct: "Monthly API credits included with the Max plan can't be used in Claude Code or the Claude apps, including for usage credits." They pay for your own apps and agents on the API. Claude Code runs on the plan's limits and, past them, on usage credits.

What happens when you hit the limit

Pro and Max have "a five-hour session limit and a weekly limit." Max and Ultra add a separate weekly limit for Fable. At a limit you have four moves:

  1. Wait for the reset: 5 hours for a session, a week for the weekly limit.
  2. Upgrade. Unused time on the old plan is credited to the new one.
  3. Turn on usage credits and keep working.
  4. Sign in with a Console account and pay per token.

What usage credits cost

Usage credits bill "at standard API rates," the table in the next section. You turn them on in Settings > Usage on claude.ai, buy a balance, and set a monthly spend limit. Running /usage-credits in Claude Code opens that page.

Two dates matter. Credits bought on or after December 7, 2026 "expire 12 months after the purchase date." Credits bought before that date don't expire.

Fast mode always draws on usage credits, even with plan usage left. It runs up to 2.5x faster at 2x the price: $16 in and $80 out per million tokens on Fable 5.2, $8 and $40 on Opus 5.5.

Claude Code API pricing and token cost

With an API key, or on usage credits, you pay these rates per million tokens:

ModelInputOutputCache read
Fable 5.2$8$40$0.20
Opus 5.5$4$20$0.20
Sonnet 5.5$2$10$0.10
Haiku 5.5, prompts up to 100K tokens$0.10$0.50$0.01
Haiku 5.5, prompts over 100K tokens$0.50$2.50$0.05

The cache-read column is the one most of a session bills at: the sample session in Anthropic's cost guide shows "91% of input tokens from cache."

How much Claude Code costs per developer per month

Anthropic publishes its own number for API billing:

Across enterprise deployments, the average cost is around $13 per developer per active day and $150-250 per developer per month, with costs remaining below $30 per active day for 90% of users.

Set that beside the subscriptions:

How you payA month
API key, Anthropic's average$150 to $250
API key, 22 active days at the $13 average22 × $13 = $286
Pro$20
Max$100
Ultra$200
Team premium seat$125

API key or subscription?

A developer who uses Claude Code every working day pays less on a subscription: Max at $100 sits under the low end of the API average, and Ultra at $200 sits inside it with no per-token meter.

An API key fits two cases. One is occasional use, a few days a month. The other is automation that no single person owns: Anthropic's GitHub Actions page says to use a Console API key for a secret shared across repositories, "since an OAuth token is tied to the subscription of the person who ran claude setup-token."

One setting decides which you are on. If ANTHROPIC_API_KEY is set in your shell, "Claude Code will use this API key for authentication instead of your Claude subscription," and you pay API rates while your plan sits unused. Run /login inside Claude Code to switch to the subscription.

Does Claude Code on the web cost extra?

No. A cloud session "counts toward your plan's usage limits alongside the rest of your Claude and Claude Code usage, and there's no separate charge for the cloud VM." Cloud sessions come with Pro, Max, Ultra, and Team plans. Sessions you run in parallel use your limits proportionately.

Code Review is the exception to "included." It "is billed on token usage through usage credits, separately from your plan's included usage," and the bill goes to the organization. A repository set to After every push multiplies that by its pushes.

What Claude Code costs in GitHub Actions

Two bills, from two companies.

Tokens, from Anthropic. With ANTHROPIC_API_KEY in the workflow, each run bills at API rates. With CLAUDE_CODE_OAUTH_TOKEN, made by claude setup-token, "runs use your Claude subscription instead of API billing." That token works on Pro, Max, Ultra, Team, and Enterprise.

Minutes, from GitHub. Anthropic's page: "the Claude Code GitHub Action runs on GitHub-hosted runners, which consume your GitHub Actions minutes."

RunnerPer minute, private repository
Linux 2-core, ubuntu-latest$0.006
Linux 4-core larger runner$0.012
Standard runner, public repositoryFree
Self-hosted runnerFree

GitHub rounds each job up to a whole minute and takes your plan's included compute off first: $12 a month on GitHub Free, $18 on GitHub Pro and GitHub Team, $300 on Enterprise Cloud.

The larger number is your own CI. Every push from a Claude Code session, local or cloud, starts your test and build workflows. A person pushes a few times a day. An agent pushes dozens of times, and three sessions in parallel push three times that. CI minutes grow with pushes, and no Claude plan covers them.

Make the CI minutes flat: Skipbox

GitHub bills nothing for a self-hosted runner's minutes. Skipbox is a self-hosted runner at a flat monthly price: $39 for 4 vCPU and 16 GB, $79 for 8 vCPU, $149 for 16 vCPU. Minutes are unlimited and nothing is metered. It runs Linux x64 jobs from public and private repositories, for organizations and personal accounts.

Your jobs also finish sooner. Each plan has a VM kept for your account, so Docker layers, npm and Cargo caches, and the checkout stay on disk between jobs. On 10 open-source projects' own CI jobs, run on GitHub's ubuntu-24.04 and on Skipbox at the same 4 vCPU, Skipbox finished the second and later runs faster on 9 of 10: ripgrep in 40 seconds against 200, astro in 30 against 528. Every run is public in the benchmarks. A faster CI is a shorter wait in every agent loop.

A job moves with one line:

jobs:
  test:
    runs-on: skipbox # was ubuntu-latest

Claude Code can make that change itself. Install the Skipbox skill with npx skills add skipbox-dev/skills, ask it to "move the Linux CI jobs to Skipbox and open a pull request," and it finds the jobs worth moving, changes their runs-on lines, and opens the pull request (Set up with an AI coding agent).

The trial takes no card. If your jobs aren't faster on Skipbox than they were on GitHub, you get every dollar back within 30 days of your first payment.

Start a free Skipbox trial on skipbox.dev

The minutes Claude's own action uses, and the settings that cap them, are in Claude Code GitHub Actions cost.

A worked month: one developer and a team of 5

Rates are the ones above. Pushes and CI length are assumptions; put your own in.

One developer building with agents

  • Claude Max, with sessions running in parallel
  • A private repository on GitHub Free, $12 of included compute
  • 50 agent pushes a day, 22 days a month
  • CI on ubuntu-latest, 10 minutes a push
LineArithmeticGitHub's runnersCI on Skipbox
Claude Max$100$100
CI minutes50 × 22 × 10 = 11,00011,000 × $0.006 = $66$0
GitHub Free's included compute−$12Untouched
Paid to GitHub$54$0
Skipbox S$39

On GitHub's runners the CI line is $54 and rises with every push. On Skipbox S it is $39, and stays there when the agents push twice as often.

A team of 5

  • Claude Team, 5 standard seats billed monthly
  • GitHub Team, $18 of included compute
  • 20 agent pushes a developer a day, 22 days a month
  • CI on a Linux 4-core larger runner, 12 minutes a push
LineArithmeticGitHub's runnersCI on Skipbox
Claude Team5 × $25$125$125
CI minutes5 × 20 × 22 × 12 = 26,40026,400 × $0.012 = $316.80$0
GitHub Team's included compute−$18Untouched
Paid to GitHub$298.80$0
3 Skipbox S plans$117

Here CI costs more than twice what Claude does: $298.80 against $125. Three Skipbox S plans run three 4 vCPU jobs at once for $117 a month, flat, and GitHub's $18 of included compute is left for Claude's own action.

Your real minutes are in GitHub's usage report. GitHub Actions pricing in November 2026 shows where to read them.

Team premium seat or Max?

Both give 5x the tier below them: a premium seat is "5x more usage than standard seats," and Max is 5x Pro. The seat is $125 a month, or $100 billed annually. Max is $100 with no annual option.

The $25 buys what a company needs: central billing and administration, spend limits an admin sets per member, and no model training on your content by default. Claude Code has come with every Team seat, standard seats included, since January 16, 2026, so a team can put premium seats on the developers who run agents all day and standard seats on everyone else.

Claude Code enterprise pricing

Anthropic's pricing page: "Pricing is $20 per seat per month plus usage billed at API rates." Enterprise is billed annually with a 20-seat minimum, and since February 12, 2026 any organization can buy it on the website with no sales call.

Usage-based Enterprise has no per-seat limit: "usage is based on consumption and billed at API rates." Budget with Anthropic's average, $150 to $250 a developer a month, and cap it with the spend limits admins set per user and per organization. SCIM, audit logs, role-based access, and custom data retention come with the plan. Volume discounts go through sales.

Where to see what you're spending

Your setupWhere the number is
Pro, Max, Ultra/usage in Claude Code, and Settings > Usage on claude.ai
Team, Enterprise/usage, and the spend report in organization analytics
API keyThe Session block of /usage, and the Console's Cost page
Bedrock, Google Cloud, FoundryYour cloud's billing console

On a subscription, /usage shows your plan's usage bars and what is using them: skills, subagents, plugins, and each MCP server, as shares of the total. With an API key, its Session block shows the session's cost, computed locally "from token counts at list price." The Console's Cost page is the bill.

What changed in 2026

Older pricing posts disagree with the table above because these changed, by the dates on Anthropic's release notes and help center:

DateChange
January 16, 2026Claude Code comes with every Team seat, standard seats included
February 12, 2026Self-serve Enterprise: one seat type, with Claude Code
September 22, 2026Opus 5.5 launches and "costs 40% less to run than Opus 5"
September 28, 2026Sonnet 5.5 launches
October 7, 2026Haiku 5.5 launches. Max and Team plans get monthly API credits
December 7, 2026Usage credits bought from this date expire after 12 months

Share what Claude Code builds

Much of what Claude Code writes for the people around you is a single HTML file: a calculator, a report, a dashboard. Drop that file on Convika's HTML upload tool and you have a link in seconds, with no account. To keep it to your team, publish it behind a password with the Pro pass, $5 once for 30 days, or Pro at $12 a month. Share a Claude Code HTML app with your team has the steps.

Frequently asked questions

How much does Claude Code cost?

$20 a month on Pro, $100 on Max, and $200 on Ultra, as of November 18, 2026. Pro is $200 if you pay yearly. Team seats are $25 a month, or $125 for a premium seat. With an API key you pay per token: $4 in and $20 out per million on Opus 5.5. Anthropic's average for API billing is $150 to $250 per developer per month.

Is Claude Code free?

No. The Free plan doesn't include Claude Code. The lowest price is Pro at $20 a month, or $200 a year.

Is Claude Pro enough for Claude Code?

Pro runs the same Claude Code and the same default model as every plan. It has a 5-hour limit and a weekly limit, shared with Claude chat, and Anthropic publishes no message count. If you hit them on larger repositories, Max gives 5x both limits for $100. Ultra gives 10x the weekly limit with no 5-hour limit on verified accounts for $200.

What is the difference between Max and Ultra?

Max is $100 a month for 5x Pro's usage every 5 hours and every week. Ultra is $200 for 10x Pro's weekly usage and no 5-hour limit on verified accounts. Both include Fable for up to half the weekly limit. Ultra is the plan that was called Max 20x.

What happens when I hit the Claude Code usage limit?

You wait for the reset, upgrade, or turn on usage credits. Usage credits bill at standard API rates and stop at the monthly spend limit you set. The session limit resets every 5 hours and the weekly limit once a week.

How much is extra usage on Claude Code?

Usage credits bill at API rates: per million tokens, $4 in and $20 out on Opus 5.5, $2 and $10 on Sonnet 5.5, $8 and $40 on Fable 5.2. Credits bought from December 7, 2026 expire 12 months after purchase.

Is it cheaper to use the API or a subscription for Claude Code?

A subscription, for daily use. Anthropic's average on the API is $150 to $250 per developer per month, above Max at $100. An API key suits occasional use and shared automation such as CI.

Can I use my Max plan's API credits in Claude Code?

No. Anthropic: "Monthly API credits included with the Max plan can't be used in Claude Code or the Claude apps, including for usage credits." They pay for your own apps on the Claude Platform.

Does Claude Code on the web cost extra?

No. A cloud session counts toward your plan's usage limits, "and there's no separate charge for the cloud VM."

Does Claude Code cost extra in GitHub Actions?

Yes, twice. Tokens bill at API rates with ANTHROPIC_API_KEY, or use your subscription with CLAUDE_CODE_OAUTH_TOKEN. GitHub bills the runner's minutes: $0.006 a minute on ubuntu-latest in a private repository. On Skipbox the minutes are flat: $39 a month, unlimited.

How much is Claude Code for a team?

Team is $25 a seat a month, or $20 billed annually, and every seat includes Claude Code. A premium seat with 5x the usage is $125, or $100 billed annually. Enterprise is $20 a seat a month, billed annually with a 20-seat minimum, plus usage at API rates.

Sources

Read 2026-11-18, Japan time.