Claude Team plan: which seat runs Claude Code

Every Team seat runs Claude Code. The help centre says it in one line: "Claude Code is included with every Team plan seat." The pricing page prices the two seats at $25 and $125 a month, $20 and $100 billed annually. What a Premium seat buys is not Claude Code but five times the usage allowance. The question for a lead is how much Claude Code each person runs in a week, and that is a figure you can read off the machines before you choose.

Claude Team plan pricing: what Anthropic publishes

From claude.com/pricing and the help centre's What is the Team plan?, read on 30 October 2026:

• Two seats. Standard, "All Claude features, plus more usage than Pro": $25 a seat a month, or $20 billed annually. Premium, "5x more usage than standard seats": $125, or $100 annually. US prices before tax; "price and plans are subject to change at Anthropic's discretion."
• 2 to 150 seats. "Team plans require a minimum of two members" and "are not intended for individual use"; they "support up to 150 seats", and above that Anthropic points at Enterprise. "Mix and match seat types": one organization holds both kinds.
• Monthly or annual. You choose at sign-up; moving from monthly to annual credits the unused time toward the first annual invoice. A seat added mid-cycle is prorated and charged immediately. A seat removed earns no immediate credit or refund; the seat is billed until the next cycle, or the annual renewal.
• Card only. The comparison table marks credit card yes, ACH and invoicing no. Single sign-on is in the plan; SCIM, audit logs, the Compliance API and custom data retention are Enterprise.
• Training. "No model training on your content by default." What Claude Code sends on each plan, and what the company's own telemetry adds about a person: Claude Code privacy.

One of Anthropic's pages lags. The billing article's worked examples price a Premium seat at $100 a month and $80 annually; the pricing page and the plan overview, read the same day, say $125 and $100. The pricing page is the one you pay on; the examples are there for the proration arithmetic.

Claude Code on the Team plan: which seat runs it

• Every seat. "Claude Code is included with every Team plan seat." The seat article lists a Standard seat as "Base features, usage limits, and Claude Code access" and a Premium seat as "Everything in Standard, plus higher usage limits". Terminal, IDE and desktop alike; a member signs in to Claude Code with "Claude account with subscription".
• What "5× more usage" means. "Team plan Standard seats include 1.25x the Pro plan's per-session usage allowance"; Premium seats "6.25x". That is the whole difference between the seats: 6.25 / 1.25 = 5.
• One allowance, per member. "Usage limits on Team plans are per-member, rather than applied to the team as a whole", and claude.ai, Claude Code and Claude Desktop "count towards the same usage limit". A long afternoon in the chat leaves less for the terminal.
• Fable. The models table puts the Fable models at "50% of weekly limits on premium seats". Opus is marked yes for the plan.
• The $500 is not for Claude Code. Each Standard seat adds $20 and each Premium seat $100 of monthly API credits to one pool capped at $500: 3 Standard and 2 Premium seats make $260, in Anthropic's own example. The credits buy the Claude API, the Console Playground, Claude Managed Agents and the Claude Agent SDK. They "can't be used in Claude Code or the Claude apps, including for extra usage", expire at the end of each billing cycle, and start after seven days on the plan.

So the seat question is the allowance. A Premium seat costs five Standard seats and holds five times the usage. A developer who works in Claude Code all day on a Standard seat meets the window; the admin's choices then are a Premium seat, usage credits, or waiting.

Claude Team plan limits: where the limit sits

Anthropic's Claude Code docs: on Team 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", its size set by the seat tier. The weekly window resets "at a fixed time each week that is assigned to your account". How high a seat's window is, in tokens or in dollars, is not published, and "usage inside the seat allowance isn't metered in dollars".

What a weekly window looks like when one person works in Claude Code all week, read every hour from one plan:

5%

Tuesday 13 October, 00:00

99%

Monday 19 October, 09:00

1%

16:00 the same day, after the reset

Six days to fill, one reset, and the next week already at 29% by Wednesday night. A seat whose holder runs Claude Code like that will meet the window every week, whatever the tier; the tier decides on which day. Caprock keeps that reading per plan, on the machine, and the chart on /teams/ draws the same file.

Past the window there are two routes:

• Wait. /usage in Claude Code shows the bars and the reset time, and Claude Code can wait for the reset and carry on with the interrupted task.
• Usage credits. Owners and Primary Owners turn them on; "usage credits are billed at standard API rates", under a monthly limit for the whole organization and one per member, and a member who reaches theirs pauses until the month ends. The spend report, open to Team plan Owners and Primary Owners, lists that spend per user and per model, refreshed daily, with a one-day delay. Inside the seat there is no dollar figure to show.

The windows, the messages and the resets, plan by plan: Claude usage limits.

Claude Team Standard vs Premium vs the API

• A Standard seat is for a developer who uses Claude Code for part of the day and the chat for the rest, inside 1.25× Pro's allowance. Who should pick it: a team whose weeks stay under the window, and whose few heavy weeks can run on usage credits.
• A Premium seat is five Standard seats' money for five times the allowance, and the only seat the Fable models are listed for. Who should pick it: the two or three people who live in the terminal, mixed in with Standard seats for the rest.
• An API key has no seat and no window, only the bill and a workspace spend cap in the Console. Who should pick it: a team that already meters by key, or one whose heaviest developer fills a Premium seat's week before Friday and would pay API rates on credits anyway.

Seats or the API: size it from measured sessions

The seat allowance has no public dollar figure. The API does, and so does every token past the allowance, since usage credits bill at the same API rates. Price the work your developers already do at API list and you have the one unit both sides share.

On my machine, Caprock priced every Claude Code session at API list rates, as Claude Code's own /usage does. Thirty days to 28 October 2026, read from that database the same day:

$7,409

thirty days, one developer, every project, at list

$13

Anthropic's average a developer an active day, across enterprise deployments

That is one developer's reading, not a typical one: $7,409 is 59 times a Premium seat's monthly price. Anthropic's own average across enterprise deployments, $150–250 a developer a month with 90% of users under $30 a day, sits above a Premium seat's price too. What a seat holds is its window, not a dollar figure, so the comparison runs through the window: as long as it does not fill, the seat is cheaper than the same tokens on a key; once it fills every week, the overflow costs what a key would.

The sizing, in three steps:

• Count merged pull requests a month, per repository. GitHub has that figure already.
• Measure what one costs at list rates on a few developers' machines, per repository, for a month. On mine the two repositories were 2.6× apart: $7.89 and $3.04 a merged pull request.
• Compare to the seats. Multiply the per-developer month at list by the headcount and set it beside Standard at $20–25 and Premium at $100–125 a seat; then read each developer's window. The ones who fill it are the Premium seats, or the usage-credit line.

What that figure is not:

• Not a bill. List rates, an estimate, as /usage is: Claude Code "computes the dollar figure locally from token counts at list price" unless an admin sets a modelPricing table. Caprock shows list.
• Not the allowance. No page states a seat's window in tokens. The curve above says when a window fills; it does not say how many dollars of work fitted inside.
• Not the chat. Caprock prices coding sessions on the machine; a chat in the browser draws on the same allowance and is not on it.

Across the team: where the figures come from

Caprock for Teams keeps the ledger across seats and API keys: every coding agent on every machine, each session priced the same way at list price, summed per repository, pull request and release, on a server you run. That is the usage line a seat plan never prints, and the number to put beside the seat menu. Buying Caprock is one flat price for the team.

From 20 seats, usage billed on top of the seat at API rates and the controls Team lacks: Claude Enterprise pricing. Pro and Max for one developer, and the API rates per token: Claude Code pricing.