"How much does Claude Code cost?" sounds like a one-line answer. It isn't, because Claude Code is billed differently depending on how you sign in. The same developer, running the same command, can draw on a subscription allowance on Monday and an API invoice on Tuesday, simply because an environment variable was set.
This guide explains the three ways Claude Code gets paid for, what happens when you hit a limit, why your plan's monthly API credits don't cover it, and how to keep spend predictable across a team.
| How you sign in | What pays for it | Where you see spend |
|---|---|---|
| Claude subscription (Pro, Max, Ultra) | Your plan's usage limits, shared with claude.ai | /usage in Claude Code; Settings → Usage on claude.ai |
| Team or Enterprise seat | Each member's seat allowance (Standard or Premium), shared with Claude chat and Cowork | Spend report in org analytics |
| Claude Console (API key) | Pay-per-token at standard API rates, billed to the organisation | Cost page in the Claude Console |
There is a fourth option for larger organisations: running Claude Code through Amazon Bedrock, Google Cloud or Microsoft Foundry, where it's billed per token to your cloud account and controlled with your cloud's budget tools.
Claude Code is included with Pro, Max and Ultra. Sign in with the same account you use for Claude and your usage draws on the plan. That usage is shared with claude.ai — chat, Claude Code in the terminal and Claude Code in your IDE all count against the same limits.
Pro and Max have a five-hour session limit and a weekly limit. Ultra has a weekly limit, and verified Ultra accounts have no five-hour limit. Max and Ultra also carry a separate weekly limit for the Fable models.
Each member's Claude Code usage draws from a per-seat allowance that resets on a rolling five-hour window and a weekly window. It's shared with Claude chat and Cowork. A Premium seat gives a heavier user more room than a Standard seat. Admins manage everything from the claude.ai admin console, not the Claude Console.
Here you pay per token, at standard API rates, with no plan limits — just your organisation's rate limits and any spend limits you set. When a developer first signs in to Claude Code with a Console account, a workspace called "Claude Code" is created automatically, so its spend is tracked separately from your other API work.
If the environment variable ANTHROPIC_API_KEY is set on a machine, Claude Code uses that key instead of your subscription, and the session is billed as API usage. This is the most common reason a team on a flat-rate plan suddenly sees an API invoice: a key left in a shell profile, a .env file or a CI runner.
To check, run /status inside Claude Code. To switch back to your plan, unset the variable and run /login.
On a subscription you'll see warnings as you approach the limit. When you reach it, you can:
/usage-credits in Claude Code to manage them. On Pro, Max and Ultra you can set a monthly spend limit.On Team and Enterprise, the seat allowance is the ceiling by default. Admins can turn on usage credits and set spend limits at organisation, group or individual level. A member without billing access who runs /usage-credits sends a request to the admins instead.
Team, Max and Ultra plans now include monthly Claude API credits — up to $500 a month on Team. They're genuinely useful, but they don't pay for interactive Claude Code in the terminal, IDE, desktop app or web, and they can't be used for usage credits either.
They only apply to Claude Code-adjacent work in two cases: claude -p and the Agent SDK, when run with an API key from the Console organisation you linked. Runs triggered by the Claude Code GitHub Action, an IDE extension or the desktop app still count as Claude Code usage. Our step-by-step guide shows how to claim your free Claude API credits and what they do cover.
Anthropic's own figure from enterprise deployments: about $13 per developer per active day, and $150–250 per developer per month, with 90% of users staying under $30 a day. Your number depends on the model, the size of the codebase and habits like running several sessions at once.
Before rolling it out widely, run a small pilot and measure. Budget more for a coding seat than a chat seat: one debugging session carries file contents, tool calls and multi-step reasoning, and can use more than a day of chat.
| Your setup | See spend | Cap spend |
|---|---|---|
| Team or Enterprise | Spend report in org analytics (CSV export) | Spend limits in admin settings |
| Claude Console | Cost page in the Console | Workspace spend limits |
| Bedrock, Google Cloud, Foundry | Your cloud billing console | Your cloud budget controls |
Habits that cut token use without slowing anyone down:
/clear between unrelated tasks. A session open all day re-sends its whole history with every request./effort.CLAUDE.md short and move specialised instructions into skills, which load only when needed./usage. On a plan it shows what's consuming your limits — subagents, MCP servers, long context — with specific recommendations.ANTHROPIC_API_KEY will put you there without you noticing.If you're setting Claude Code up for a team and want the sign-in, seats, spend limits and data handling right first time — including the POPIA questions about what code and data leave your environment — book a Free AI Assessment. For the wider Claude plan picture, see our Claude Team billing guide.
Next step
Our free AI assessment is a scoped conversation about your systems, your constraints and what is actually worth automating — not a product demo. You leave with a plan you can act on, whether or not you go on to work with us.
Get a Free AI Assessment
Claude Team, Max and Ultra plans include up to $500 a month in API credits that don't appear in claude.ai. Who qualifies, how to claim them, and what they cover.
Comparing Claude Team vs. Enterprise pricing for South African businesses in 2026. Understand costs, features, and when to upgrade for your AI automation needs.
Looking for a Copilot Studio implementation partner in South Africa? We cut through the hype to show you what matters: real pipelines, not just decks.