Recipe · Basic · 5 min
Is Claude Code free? Pricing and which plan to pick (2026)
Claude Code isn't free: it comes with Pro, Max and Ultra, or you pay per token through the API. Official prices, which plan to pick and how to spend less.
Tested with Claude Code 2.1.316 · reviewed
The problem
You want to use Claude Code, but you're not sure whether it's free, what it costs or which plan suits you. The web is full of outdated, contradictory figures, and country searches ("claude code price mexico", "argentina", "spain"...) turn up prices nobody checks. This page only uses what Anthropic's official pages say today, with the date they were checked.
The solution
1. The short answer: no, Claude Code isn't free
Claude's Free plan ($0) covers chat on web, desktop and mobile, but it doesn't include Claude Code. On the pricing page, the "Claude Code included" line starts at Pro (claude.com/pricing, checked 31-Oct-2026).
Installing the program costs nothing, but to use it you need one of these two things:
- A subscription that includes it (Pro, Max, Ultra, Team or Enterprise): you pay a fixed fee and get a usage allowance that renews.
- An API key (Claude Console account): you pay for every token it uses.
2. Subscription plans that include Claude Code
Prices in US dollars, excluding tax (claude.com/pricing, checked 31-Oct-2026):
| Plan | Price | Included usage |
|---|---|---|
| Free | $0 | No Claude Code |
| Pro | $20 a month, or $17 a month billed annually ($200 up front) | "More usage than Free", every 5 hours and every week |
| Max | $100 a month | 5x Pro usage, every 5 hours and every week |
| Ultra | $200 a month | 10x Pro usage per week and no 5-hour limit (verified accounts) |
| Team, standard seat | $20 per seat a month billed annually, $25 billed monthly | For teams of 2 to 150 |
| Team, premium seat | $100 per seat a month billed annually, $125 billed monthly | 5x the usage of a standard seat |
| Enterprise | US$20 per seat a month billed annually, 20-seat minimum, plus usage at API rates | Depends on model and task |
How subscription limits work (help center, checked 31-Oct-2026):
- Pro and Max have a five-hour session limit and a weekly limit. Ultra has the weekly one, and verified Ultra accounts don't have the five-hour limit.
- The limits are shared between Claude chat and Claude Code: everything you do in both counts against the same allowance.
- Anthropic doesn't publish a fixed number of messages per five hours. It depends on the model, the size of your project and how long your sessions are.
When you hit the limit, you can wait for the reset or turn on usage credits: they're prepaid, billed at API rates, and you can set a monthly cap (help center, checked 31-Oct-2026). From Claude Code, open them with /usage-credits.
3. Pay as you go with the API
With a Claude Console account you pay per million tokens (MTok). List prices (claude.com/pricing, checked 31-Oct-2026):
| Model | Input | Output |
|---|---|---|
| Opus 5.5 | $4 / MTok | $20 / MTok |
| Sonnet 5.5 | $2 / MTok | $10 / MTok |
| Haiku 5.5 (prompts up to 100K tokens) | $0.10 / MTok | $0.50 / MTok |
| Haiku 5.5 (prompts over 100K tokens) | $0.50 / MTok | $2.50 / MTok |
What does that mean in practice? The official docs give this figure from enterprise deployments: around $13 per developer per active day and $150-250 per developer per month, with 90% of users staying below $30 per active day (costs docs, checked 31-Oct-2026). Your spend can be very different: it changes with the model, the size of the codebase and what you automate.
To pay through the API, start claude, type /login and pick the Console account. If the ANTHROPIC_API_KEY variable is set, Claude Code uses that key instead of your subscription and bills you through the API, even if you pay for Pro or Max.
4. Which plan to pick
Pick by how much you use it, not by the lowest price:
- You want to try it or use it a few hours a week: Pro. It's the cheapest way into Claude Code.
- You use it every day and keep hitting the five-hour limit: Max.
- You run very long sessions or tasks that take hours: Ultra (no five-hour limit on verified accounts).
- A team of 2 to 150 people with shared billing: Team (standard or premium seat depending on each person's usage).
- Occasional use, scripts, CI, or you want to pay only for what you use: API. Set a spend limit on the Console workspace: with the API there's no ceiling except the one you set.
If you're unsure, start with Pro billed monthly and watch /usage for a week or two before upgrading.
5. How to see your usage
Type these commands inside a Claude Code session (commands reference, checked 31-Oct-2026):
| Command | What it shows |
|---|---|
/usage | Session cost, your plan's usage limits and activity stats. On paid plans, what uses up your allowance (skills, subagents, MCP) with recommendations |
/cost | Alias for /usage |
/stats | Alias for /usage, opens on the Stats tab |
/status | Version, model and the account you signed in with (subscription or API) |
/context | What fills your context window right now |
/insights | HTML report on how you work, with suggestions |
Two details from the costs docs:
- In
/usage, the session's dollar figure is an estimate at list price, meant for API users. On Pro, Max or Ultra it isn't what you're charged: look at the plan usage bars. - If the
/usagefooter showsf to fix, pressfand Claude Code applies the recommendations it can (version 2.1.305 or later).
6. How to spend less
The tips from the official docs that save the most:
/clearbetween unrelated tasks. Claude Code resends the whole conversation with every message, so old context gets paid for again and again./clearcosts nothing;/compactdoes, because it has to read the conversation to summarize it.- Pick the right model. Sonnet handles most coding tasks and costs less than Opus. Keep Opus for architecture decisions or multi-step reasoning. Switch with
/model, and setmodel: haikuon simple subagents. - Lower the effort on simple tasks with
/effort(for example/effort low). Thinking tokens are billed as output tokens. - Short CLAUDE.md. It loads in full at the start of every session. The docs recommend keeping it under 200 lines and moving workflow-specific instructions into skills, which only load when used.
- Disable MCP servers you don't use with
/mcp, and prefer command-line tools likeghwhen they exist. - Ask for specific things. "Improve this codebase" makes Claude read half the project; "add input validation to the login function in auth.ts" doesn't.
- Use plan mode (
Shift+Tab) for big tasks, and pressEscas soon as you see it heading the wrong way.
Check it
Open Claude Code in your project and see which account you're paying with:
claude --version
claude
Inside the session:
/status
/usage
/status shows the account you signed in with. In /usage, on a subscription you'll see the usage bars for the five-hour and weekly windows; on the API, the session's estimated cost.
To find out whether an API key is overriding your subscription without printing the key:
[ -n "$ANTHROPIC_API_KEY" ] && echo "API key found: Claude Code will bill through the API" || echo "No API key: your subscription is used"
Why it works
Claude Code always uses model tokens. What changes is who pays for them. On a subscription, tokens come out of your plan's allowance, which renews every five hours and every week. On the API, every token is billed at list price. That's why the same tips (small context, right model) work in both cases: on a subscription you take longer to hit the limit, and on the API you pay less.
Official sources, checked 31-Oct-2026: Claude pricing, manage costs in Claude Code, Claude Code with Pro or Max, usage credits, the Pro plan and Pro and Max invoices.
Common mistakes
- You pay for Pro or Max and still get an API bill →
ANTHROPIC_API_KEYis set and Claude Code uses it instead of your subscription → remove it from your~/.bashrcor~/.zshrc, open a new terminal and sign in again with/login. - "You've hit your session limit" mid-task → the five-hour window's allowance ran out, and it's shared across all models → wait for the reset shown in the message, turn on usage credits with
/usage-credits, or upgrade. Switching models with/modeldoesn't fix it. - You hit the limit very fast → a session left open all day resends its whole history with every message → use
/clearwhen you switch tasks and check in/usagewhat uses the most. - The
/usagecost doesn't match what you pay → it's a local estimate at list price → on a subscription, look at the plan bars; on the API, the real bill is on the Claude Console usage page.
FAQ
Is there any official way to use Claude Code for free?
No. The Free plan doesn't include Claude Code and the API charges per token. The program installs for free, but using it needs a paid subscription or API credit.
How much does Claude Code cost in my country?
Anthropic publishes prices in US dollars (USD), excluding tax. According to its help center, you're charged in your local currency where supported, monthly pricing varies by region, and tax is calculated from your payment method's billing address: some regions include it in the displayed price and others add it at checkout. You'll see the exact amount for your country on the checkout screen before you confirm. We don't list prices by country because Anthropic doesn't publish one.
Is Pro enough for coding?
For trying Claude Code and a few hours a week, yes. If you work with it daily and keep hitting the five-hour limit, the next step is Max (5x Pro usage).
Are chat and Claude Code limits separate?
No. They're the same: everything you do in Claude chat and Claude Code counts against the same allowance.
Which is better, a subscription or the API?
If you use it often, a subscription, because the price is fixed and predictable. If you use it rarely or in scripts and CI, the API, because you only pay for what you use. In that case, set a spend limit in Console.
Can prices change?
Yes. Anthropic's own page says prices and plans may change at its discretion. That's why this page carries a review date; check claude.com/pricing before you pay.