Claude Code for teams: cost per developer and first settings

The cost of Claude Code for teams depends on how developers sign in: with a seat on Anthropic’s Team plan, where Claude Code is included, or with a Claude Console account billed per token.

In short:

Claude Code is included with every seat on Anthropic’s Team plan, which is for teams of 2 to 150. A Standard seat costs $20 a month billed annually or $25 monthly. A Premium seat costs $100 or $125 and gives 5 times the usage of a Standard seat.
Each developer works from a seat allowance that resets every five hours and every week, and it’s shared with Claude chat and Cowork. Past the allowance, usage credits bill at standard API rates, up to spend limits you set.
If you pay through the API instead, Anthropic’s averages across enterprise deployments are about $13 per developer per active day and $150 to $250 per developer per month, with 90% of users under $30 per active day.
For an 8-person team with 6 Standard and 2 Premium seats on annual billing, that’s $320 a month in seats. The same 8 developers at Anthropic’s API averages would cost $1,200 to $2,000.
In the first week, set managed settings that block reads of secret files and turn off bypass mode, pick a default model, and put a monthly cap on usage credits.

All prices below were checked on Anthropic’s pages on 12 November 2026. They change, so check them again before you budget.

What does Claude Code cost on a Team plan?

Anthropic’s pricing page lists two seat types on the Team plan:

Seat	Billed annually	Billed monthly	Usage
Standard	$20 per seat a month	$25 per seat a month	More than an individual Pro plan
Premium	$100 per seat a month	$125 per seat a month	5 times a Standard seat

You can mix the two on one plan. There is no separate Claude Code seat: Anthropic’s help article on Claude Code for teams says “Claude Code is included with every Team plan seat.”

Each seat comes with a usage allowance. Anthropic’s Claude Code costs page explains that on Team and Enterprise plans “each member’s Claude Code usage draws from a per-seat allowance that resets on a rolling five-hour window and a weekly window.” The same allowance covers Claude chat and Cowork, so a developer who also uses chat heavily has less left for coding. Anthropic doesn’t publish the allowance in tokens or dollars, only relative to other plans.

When a developer runs out, they wait for the window to reset, or you turn on usage credits. According to Anthropic’s usage credits article for Team plans:

“Usage credits are billed at standard API rates.”
An Owner or Primary Owner turns them on and can set a monthly spend limit for the whole organization.
“Spend limits for individual members aren’t available to every Team organization.” Check your Usage page to see whether yours has them.
When someone reaches their limit, their usage credits pause. Limits are checked before each request, so one request can take spend slightly over the limit before further requests are blocked.

The Team plan also lists up to $500 a month in pooled API credits, but Anthropic’s Claude Code for teams article says “Monthly API credits included with Team plans can’t be used in Claude Code or the Claude apps, including for extra usage.”

If you’re also choosing a chat assistant for the rest of the company, our comparison of Claude Team and ChatGPT Business covers that side.

What does Claude Code cost through the API?

Developers can also sign in with a Claude Console account and pay per token. The costs page gives Anthropic’s averages from enterprise deployments: “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.” Those are averages from larger companies, so treat them as a starting estimate. Anthropic’s own advice is to run a small pilot and measure before a wider rollout.

The per-token prices are on Anthropic’s API pricing page. Per million tokens, Claude Sonnet 5.5 costs $2 for input and $10 for output, and Claude Opus 5.5 costs $4 and $20.

On API billing your controls are different. You cap spend with workspace spend limits in the Console, where Claude Code gets its own workspace the first time someone signs in. The costs page also recommends rate limits per user by team size:

Team size	Tokens per minute, per user	Requests per minute, per user
1 to 5 users	200k to 300k	5 to 7
5 to 20 users	100k to 150k	2.5 to 3.5
20 to 50 users	50k to 75k	1.25 to 1.75

The per-user figure falls as the team grows because fewer people use Claude Code at the same moment.

Standard or Premium seats: how do you decide?

On annual billing a Premium seat costs 5 times a Standard seat and gives 5 times the usage, so the price per unit of usage is the same. The decision is about which developers need the larger allowance.

A worked example for 8 developers, using the prices above:

Setup	Cost a month
8 Standard seats, annual	8 × $20 = $160
6 Standard and 2 Premium, annual	6 × $20 + 2 × $100 = $320
6 Standard and 2 Premium, monthly	6 × $25 + 2 × $125 = $400
8 developers on API billing at Anthropic’s averages	8 × $150 to $250 = $1,200 to $2,000

Seat costs are fixed. Usage credits come on top at API rates, up to whatever cap you set.

A simple way to decide:

Start everyone on Standard.
Turn on usage credits with an organization cap you could absorb in a heavy month.
After a month, look at who hit session or weekly limits and who spent usage credits.
A Premium seat costs $80 a month more than a Standard seat on annual billing. If a developer’s usage credits regularly come to more than that, try them on Premium.

How do you track Claude Code spend?

On a Team plan you have three places to look.

/usage inside Claude Code. On a subscription it shows plan usage bars and a breakdown of what used the allowance, such as skills, subagents and MCP servers, with suggested changes. Press d or w to switch between the last 24 hours and 7 days. The figures come from session history on that machine, so they leave out other devices and claude.ai.
The spend report. Team Owners and Primary Owners can export it as a CSV from Settings > Analytics, according to Anthropic’s usage analytics article. Each row is one person’s use of one model, with request counts, tokens and spend in US dollars. It refreshes daily with a one-day delay and goes back up to 90 days. It shows usage-credit spend only, because usage inside the seat allowance isn’t metered in dollars.
OpenTelemetry. Setting CLAUDE_CODE_ENABLE_TELEMETRY=1 sends metrics such as claude_code.cost.usage and claude_code.token.usage to your own monitoring. The monitoring docs show how to set the variables for everyone through managed settings.

For a pilot, the spend report and a weekly look at who hit limits are enough. Add OpenTelemetry once you have more than a handful of developers or need per-person numbers in your own dashboards.

Which settings should you set in the first week?

Claude Code reads settings from several files, and managed settings sit above all of them. No developer’s user, project or local settings can override a managed key. On Team and Enterprise plans an Owner or Primary Owner can set them in the claude.ai admin console under Organization settings > Claude Code > Managed settings. Without the console, you can place a managed-settings.json file on each machine instead, at /Library/Application Support/ClaudeCode/ on macOS, /etc/claude-code/ on Linux and WSL, or C:\Program Files\ClaudeCode\ on Windows.

A small first policy, based on Anthropic’s example:

{
  "permissions": {
    "deny": [
      "Read(./.env)",
      "Read(./secrets/**)"
    ],
    "disableBypassPermissionsMode": "disable"
  }
}

The deny rules stop Claude Code’s file tools, and file commands it recognises in the shell such as cat, from reading secret files. The docs warn that a command that reads files without naming them can still get through, so keep real secrets out of repositories anyway. Per the permissions docs, rules are checked in the order deny, ask, allow, so an allow rule elsewhere can’t override a deny. disableBypassPermissionsMode stops anyone switching to the mode that skips permission prompts. If you want managed settings to be the only source of permission rules, the docs also cover allowManagedPermissionRulesOnly, though that blocks the project-level allow rules your developers would otherwise share.

The rest of the first-week list:

Default model. A managed model sets the model new sessions start on, and developers can still switch. To restrict which models they can use, set availableModels. The settings docs cover both.
Shared project settings. Commit .claude/settings.json to each repository with allow rules for safe, routine commands such as your lint and test scripts, so developers see fewer prompts.
Usage credits. In Organization settings > Usage, turn on usage credits, set the organization’s monthly limit, and set per-member limits if your plan has them.
A written policy. Settings control what the tool can do. People also need to know what they may paste into it. Our one-page AI acceptable use policy template is a starting point.
Check it worked. Run /status on one developer’s machine. The Setting sources line shows whether managed settings reached it.

Which habits cut usage?

These come from Anthropic’s costs page. Each one stretches a seat allowance further, or cuts the bill on API billing.

Clear between tasks. /clear starts fresh when you switch to unrelated work. Old context is sent again with every message, so a long session costs more per question. Use /rename before clearing if you want to come back with /resume.
Match the model to the job. Anthropic says Sonnet handles most coding tasks and costs less than Opus, and suggests keeping Opus for complex architecture and multi-step reasoning.
Lower the effort level for simple tasks with /effort. Thinking can’t be turned off on Opus 5.5, Sonnet 5.5 or Haiku 5.5, so effort is the control to use on those models.
Keep CLAUDE.md short. It loads at the start of every session, and Anthropic suggests keeping it under 200 lines and moving workflow-specific instructions into skills, which load only when used.
Turn off MCP servers you aren’t using, with /mcp.
Write specific prompts. “Add input validation to the login function in auth.ts” reads fewer files than “improve this codebase.”

Rolling it out to your developers

We set up Claude, ChatGPT or Copilot for teams and help developers adopt Claude Code. Our team rollout prices are fixed: a half-day workshop is $1,500, a 2-day rollout is $2,500 and a 2-week rollout is $7,500. Seats are billed to your own Anthropic account at Anthropic’s price, and we add no markup.

To see what each rollout format covers, read the team rollout service page.