Claude Code Pricing Explained: Plans, Credits and Plan Mode

Every number on this page was read off Anthropic's own pricing page and Claude Code documentation on 6 September 2026. Anthropic states plainly that price and plans are subject to change at its discretion, and Claude Code ships new versions constantly. Treat this as a snapshot and a map of where to look, not a quote. The live page is claude.com/pricing.

Overview

Claude Code has no price of its own. It is bundled into paid Claude subscriptions, and it can also be run against pre-paid API credits through a Claude Console account. Those are the only two billing paths, and which one you are on decides what your limits look like, where you see your usage, and what happens when you run out.

This post covers both, then explains plan mode, which people search for right next to "claude code pricing" and which has nothing to do with billing.

The short version

• The free Claude.ai plan does not include Claude Code.
• Every paid plan does. There is no separate Claude Code purchase.
• Your terminal work and your Claude chats draw from the same usage pool.
• Limits reset on a rolling five-hour window, and paid plans add weekly limits on top.
• If you prefer paying per token instead, sign in with a Console account.

1. Two Ways Claude Code Is Billed

Subscription. You pay a flat monthly or annual fee for a Claude plan, and Claude Code is one of the things it includes. Anthropic's pricing FAQ states it directly: Claude Code is included in all paid plans, and it shares the same usage limits as the rest of your plan, so terminal work and chats draw from one pool. There is no per-token bill.

API credits. You create a Claude Console account, load pre-paid credits, and sign Claude Code in with that. Every token is then metered. On first login, Anthropic creates a "Claude Code" workspace in the Console so the spend is tracked in one place.

Run /login to move between them. Watch for one thing: if ANTHROPIC_API_KEY is set in your shell, Claude Code uses that key rather than your subscription, and in non-interactive -p mode it always does. An old key in a .zshrc will quietly bill the wrong account. /status shows which credential is in use.

2. What Each Plan Includes, as Published Today

Read off claude.com/pricing on 6 September 2026. Prices exclude tax.

Free, $0. The comparison table lists Claude Code as No on this plan, along with Claude Cowork, Claude Design, and Claude Science. Sonnet is available, Opus is not. If your search was "is Claude Code free", that is the answer.

Pro, $17 per month with the annual discount ($200 billed up front), or $20 per month billed monthly. Includes Claude Code, more usage than Free, and access to Opus.

Max, from $100 per month. Everything in Pro, plus a choice of 5x or 20x more usage than Pro per five-hour session, higher output limits, early access to some features, and priority access at busy times. Max is billed monthly only. The comparison table has separate Max 5x and Max 20x columns, but the card itself only prints "From $100". I could not find the Max 20x figure stated as a number anywhere on the pricing page, so I am not going to invent one. Open the plan picker on claude.com/pricing and it will show you both.

Team, for 2 to 150 people. Two seat types. A Standard seat is $20 per seat per month billed annually, or $25 monthly. A Premium seat is $100 per seat per month billed annually, or $125 monthly, and Anthropic describes it as 5x more usage than a Standard seat. You can mix seat types in one organisation.

One honest caveat here. The pricing page prints "Includes Claude Code and Claude Cowork" as a bullet immediately under the Premium seat block, while the same page's FAQ says Claude Code is included in all paid plans. Those two statements are not obviously consistent for a Team Standard seat. If you are buying Team seats specifically to get Claude Code, confirm with Anthropic which seat type carries it before you commit.

Enterprise, $20 per seat per month plus usage billed at API rates. Billed annually. Everything in Team plus SCIM, audit logs, custom data retention, role-based access, IP allowlisting, and org-level spend controls.

3. What the API Path Costs

If you sign in with a Console account, you pay published token rates. From the API tab of the pricing page on 6 September 2026, per million tokens:

Model	Input	Output
Fable 5.1	$10	$50
Opus 5	$5	$25
Sonnet 5	$2	$10
Haiku 4.5	$1	$5

Prompt caching is billed separately, and cache reads are far cheaper than fresh input. Sonnet 5 listed a cache read at $0.20 per million tokens against $2 for input, which is why caching matters so much to an agent that re-sends a large context every turn. US-only inference is available at 1.1x the input and output rates.

Anthropic's own cost documentation says that across enterprise deployments the average is around $13 per developer per active day and $150 to $250 per developer per month, with 90% of users below $30 per active day. Those are Anthropic's published figures for team rollouts, not a benchmark I ran, and a solo builder on a small repo will look nothing like the average of a hundred-engineer deployment. Treat it as an order of magnitude.

4. Plan Mode Is a Feature, Not a Price Tier

"Claude code plan mode" gets searched right alongside "claude code pricing", and I suspect some of those people are hunting for a cheap tier. It is not that. Plan mode is one of six permission modes, and a permission mode decides what Claude can do without asking you:

Mode	Runs without asking	Suited to
default (shown as Manual)	Reads only	Reviewing every action, sensitive work
acceptEdits	Reads, file edits, common filesystem commands	Iterating on code you are watching
plan	Reads, plus exploration commands	Understanding a codebase before changing it
auto	Everything, with background safety checks	Long tasks, fewer prompts
dontAsk	Only pre-approved tools	Locked-down CI
bypassPermissions	Everything	Isolated containers and VMs only

In plan mode, Claude reads files, runs commands to explore, and writes a plan. It does not edit your source. Edits stay blocked until you approve the plan. When the plan is ready you get three choices: approve and switch to auto mode, approve and review each edit yourself, or keep planning and tell it what to change. Ctrl+G opens the proposed plan in your editor so you can rewrite it before Claude proceeds.

Three ways in:

claude --permission-mode plan

Or press Shift+Tab inside a session until the status bar reads ⏸ plan mode on. Or prefix a single prompt with /plan. To make it the default for a project, set defaultMode to plan in .claude/settings.json.

Where it touches cost: plan mode still sends requests, so exploring a large codebase draws on the same usage pool as editing it. It is not a free tier. What it buys you is control before the diff. On Pro, Max, and Team plans, interactive terminal sessions start in auto mode by default, so if you want plan mode you have to ask for it. Worth knowing before your first session on real code, which the install guide walks through.

5. How the Usage Limits Work

Anthropic does not publish a message count, and says why: how much you can do depends on conversation length and complexity, the model, and the features you use. What is documented:

• Every plan has limits that reset on a rolling five-hour session window.
• Paid plans add weekly limits on top.
• Web, desktop, mobile, and Claude Code all draw from one pool.
• Pro gives at least 5x more per five-hour session than Free. Max gives 5x or 20x more than Pro. On Team, Standard seats give more than Pro and Premium seats give 5x more than Standard.
• Anthropic reserves the right to add further caps, including weekly and monthly ones, at its discretion.

When you hit a limit you have three options: wait for the window to reset, move to a higher plan, or turn on usage credits, which let paid plans keep working at standard API rates past the plan limit. Run /rate-limit-options inside a session to see them, or /usage-credits to configure credits and a monthly spend limit. On Pro and Max that opens Settings then Usage on claude.ai. On Team and Enterprise it opens your org's admin settings, or sends a request to an admin if you lack billing access.

Claude Code can also just wait. The status line shows Usage limit reached · continuing automatically at 3:45pm, and the autoContinueAtUsageLimit setting controls whether it does that.

6. Seeing Your Own Usage

Run /usage in a session. /cost and /stats are aliases.

On a subscription you get plan usage bars, activity stats, and a breakdown of what consumed the usage: skills, subagents, plugins, and individual MCP servers, each as a percentage. It flags behaviours accounting for 10% or more of recent usage, such as long context or cache misses. Press d or w to switch between the last 24 hours and the last 7 days.

Two limits on that number. It is computed from local session history on that machine, so another laptop or claude.ai is not counted. And the Session block's dollar figure is computed locally at list price, which matters to API users and is largely irrelevant on a subscription. For API billing, the authoritative record is the Console usage page.

For per-run caps in scripts, claude -p accepts a hard budget:

claude -p --max-budget-usd 5.00 "run the test suite and fix what fails"

7. Which One Should a Solo Builder Pick

I am not going to tell you to spend more. Here are the actual trade-offs. The same logic I use for choosing a first automation tool applies: start on the cheapest thing that works and let real usage make the next decision.

Free is not an option if the tool you want is Claude Code. That is not a nudge, it is what the comparison table says.

Pro is the smallest thing that works. You get Claude Code, Opus access, and the rest of the paid surface. The trade-off is the ceiling: one long agentic session on a large repository can move through a five-hour window quickly, and you share that window with your chats. Annual saves money and locks you in for a year on a product that changes every week, which is a real cost in a market this young.

Max buys headroom, not features. The advertised difference is usage, output limits, early access, and priority at busy times. It is worth it only if you are actually hitting Pro's limits, and /usage will tell you. Watching the bars for a month is cheaper than guessing.

Usage credits are the middle path most people miss. Instead of jumping a tier for two busy weeks, turn credits on and pay API rates for the overflow, under a monthly spend limit you set. If your heavy weeks are occasional, this is likely cheaper than Max.

A Console account suits spiky, scriptable work. If most of your usage is claude -p in CI rather than interactive sessions, per-token billing with --max-budget-usd caps is easier to reason about than a shared window. If you sit at a terminal all day, a subscription is simpler and has no bill surprise.

The honest answer for someone starting out: take Pro monthly, use it for a month, look at /usage, and only then decide whether the answer is Max, credits, or nothing.

8. What I Could Not Verify

Stating these plainly is more useful than filling the gaps:

• The Max 20x monthly price. The pricing page card says "From $100" and the comparison table lists Max 5x and Max 20x as separate columns without printing the second figure. Check the plan picker.
• Whether a Team Standard seat includes Claude Code. The page's FAQ says all paid plans include it; the seat bullets put it under Premium. Confirm before buying seats for that reason.
• Regional pricing and tax. All figures above exclude applicable tax and reflect the page as served to me.
• Exact token allowances per window. Anthropic deliberately does not publish these, and anyone quoting you a message count is guessing.

Further reading

• Pricing: https://claude.com/pricing
• Manage costs effectively: https://code.claude.com/docs/en/costs
• Choose a permission mode, including plan mode: https://code.claude.com/docs/en/permission-modes
• Slash command reference: https://code.claude.com/docs/en/commands

---

Next step

If you have not installed it yet, start with How to Install Claude Code, which covers macOS, Linux, Windows, WSL, and the VS Code and JetBrains integrations.

Once it runs, the free Claude Code for Builders course in the House of Loops classroom walks permission modes and a first real project end to end: open the classroom. Or join the free weekly list for one working build a week. More about who is behind this on the about page.