In short
Claude Code is included with Claude Pro ($20 a month, or $17 billed annually), Max ($100), Ultra ($200) and Team and Enterprise seats, or you can pay per token through the API. Anthropic reports enterprise averages of around $13 per developer per active day and $150–$250 a month. Subscriptions suit regular use. Long sessions, Opus use and many subagents burn the most, so manage context carefully.
The two ways to pay
Subscription. You pay a flat monthly fee, and usage is limited by rolling windows.
API. You pay for the tokens you use, through a Claude Console account or a cloud provider such as Amazon Bedrock. There's no fixed ceiling unless you set one.
Prices are from Claude's pricing page and API pricing, checked on 4 November 2026. This article relies on Anthropic's own documentation, so it's a product explanation, not an independent test.
| Plan | Price | Notes |
|---|---|---|
| Free | $0 | Claude Code not included |
| Pro | $20/month; $17 billed annually | Good for regular but moderate use |
| Max | $100/month | Higher limits |
| Ultra | $200/month | Higher limits still |
| Team | Standard $20/seat billed annually ($25 monthly); Premium $100 ($125 monthly) | 2–150 seats; choose the tier by usage |
| Enterprise | $20/seat billed annually, 20-seat minimum, plus usage at API rates | Admin controls |
What real usage costs
Anthropic's cost documentation provides some useful figures: across enterprise deployments, the average is around $13 per developer per active day, and $150–$250 per developer per month, with costs staying below $30 per active day for 90% of users.
Treat that as a guide, not a quote. Anthropic notes that costs vary widely by model, codebase size and whether you run multiple instances or automation. It suggests starting with a small pilot group to establish your own baseline.
For API billing, current rates per million tokens include:
| Model | Input | Cache read | Output |
|---|---|---|---|
| Claude Haiku 5.5 (prompts up to 100K tokens) | $0.10 | $0.01 | $0.50 |
| Claude Sonnet 5.5 | $2 | $0.10 | $10 |
| Claude Opus 5.5 | $4 | $0.20 | $20 |
| Claude Fable 5.1 | $10 | $0.25 | $50 |
How subscription limits work
On Team and Enterprise, each member's allowance resets on a rolling five-hour window and a weekly window, and it's shared with Claude chat and Cowork. That sharing matters: heavy coding leaves less for everything else. When you hit a session or weekly limit, the message tells you when it resets, and switching model usually won't help with those. You can also enable usage credits, to continue beyond your allowance, at extra cost.
Inside Claude Code, run /usage. On a subscription, it shows plan usage bars, plus a breakdown of what's using your allowance, such as skills, subagents and MCP servers, with suggestions. On an API plan, it shows token counts and an estimated cost.
Subscription or API?
- Coding most days: a subscription is usually better value, and simpler to budget for.
- Occasional use: Pro is cheap, or use the API with a low monthly limit.
- Automation in CI: API billing with workspace limits gives you clearer control, although check the current options for your plan.
- A team with mixed usage: give heavy users Premium seats and others Standard, rather than the same tier for everyone.
- Data residency or a cloud commitment: use a cloud provider, and control spend through your cloud's budget tools.
If you're on the API, compare against your bill after a typical month. If you regularly spend above a subscription's price, it's worth switching.
Why usage climbs
Anthropic's docs explain a point that surprises many people: Claude Code sends your whole conversation with every request. So a one-line question in a session that's been open all day still draws usage for the entire conversation, at cached rates where the cache is warm.
Things that drive usage up:
- Long sessions that run across unrelated tasks
- Cache misses, when you come back after a break longer than the cache lifetime
- Subagents and agent teams. Anthropic says agent teams can use about 7x more tokens than standard sessions, when teammates run in plan mode
- Scheduled tasks and loops, which send your context each time they fire
- Expensive models used for simple jobs
- Large CLAUDE.md files, loaded at the start of every session
How to reduce it
- Run
/clearbetween unrelated tasks. Old context costs tokens with every message. It's probably the single largest saving for most people. - Match the model to the job. Anthropic says Sonnet handles most coding tasks well, so reserve Opus for complex reasoning, and use a smaller model for simple subagent work.
- Keep your CLAUDE.md short. Anthropic suggests under 200 lines, moving detailed workflows into skills, which load only when needed. See how to write a CLAUDE.md file.
- Be specific. “Add input validation to the login function in auth.ts” costs much less than “improve this codebase”, as it avoids broad scanning.
- Use plan mode on complex tasks, so that you can stop a wrong direction before it costs you.
- Stop early. Press Escape when it's going the wrong way, rather than letting it finish.
- Lower the effort level on simple tasks.
- Prefer CLI tools over MCP servers where possible, and disable servers that you aren't using.
- Offload verbose work. Delegate test runs and log processing to subagents, or filter output with hooks, so the main conversation stays small.
Budgeting for a team
- Begin with a pilot of a few developers, and measure for a month.
- Budget more for coding seats than chat seats, as Anthropic recommends.
- Set spend limits: at the organisation, group and member level on Team and Enterprise, or by workspace in the Console.
- Review the spend report and analytics monthly, and share good habits between developers.
- Be careful with parallel agents. They can be productive, but costs scale with their number.
And remember what you're comparing against. If a tool saves a developer a couple of hours a week, then its monthly cost is modest. It's the undisciplined usage that wastes money, not the tool itself. See also what vibe coding costs per month and how it compares with Codex.
Want help setting up AI-assisted development with sensible limits? Book a call.
Common questions
Is Claude Code free?
No. It isn't included on Claude's Free plan. It comes with paid plans starting at Pro, at $20 a month or $17 billed annually, or you can use it through API billing and pay for the tokens you use.
How much does Claude Code cost per month?
Anthropic reports that across enterprise deployments, the average is around $150–$250 per developer per month, or about $13 per active day. Individuals on a Pro plan pay a flat $20. Your costs will depend on models, codebase size and usage patterns.
Why did I hit my Claude Code limit so quickly?
Long sessions, expensive models, subagents and large context can all consume your allowance fast, because the full conversation is sent with each request. Use /clear between tasks, choose a suitable model, and check /usage to see what's using it.
Is the Max plan worth it for Claude Code?
It depends on whether you regularly hit Pro's limits. If you code with it most days and keep running out, Max is often cheaper than API billing at the same usage. Try Pro first, and upgrade if you hit limits frequently.