Deployment Labs
Deployment NotesCost control

Claude Enterprise Pricing in 2026: Seats, Usage and the Team Plan Compared

Deployment LabsUpdated 13 min read

Claude Enterprise and a Claude Team seat have the same list price: $20 a user a month on annual billing. That makes Enterprise look like a free upgrade. It isn't. A Team seat comes with a usage allowance. An Enterprise seat comes with none, and every token your staff use is billed on top at API rates. At list price, Enterprise never costs less than standard Team seats on annual billing for the same people. Companies buy it for the controls, the scale or the compliance, and they need a budget for usage that the price page can't give them.

How much does Claude Enterprise cost?

Claude Enterprise costs $20 per seat per month, billed annually, with a 20-seat minimum, plus all usage billed at Anthropic's API rates, as of November 2026. Anthropic's pricing page puts it as "Seat price + usage at API rates" and adds: "Usage cost scales with model and task." Its billing guide is blunter: the seat "doesn't include any usage. Every token your team consumes is billed separately at standard API rates."

This guide is for the people who sign off an AI budget in a company of 20 to 1,000 people: COOs, finance leads, IT leads and whoever runs the rollout. It covers what Anthropic publishes for Claude Team and Claude Enterprise, how the usage side of the bill works, where each plan costs less, and how to budget. All prices are Anthropic's published US prices as of November 2026, before tax.

Plan Seat price Usage Size
Claude Team, standard seat $20 a month annual, $25 monthly Included, up to the seat's limits 2 to 150 people
Claude Team, premium seat $100 a month annual, $125 monthly Included, 5 times a standard seat 2 to 150 people
Claude Enterprise, self-serve $20 a month, annual only None included; all at API rates 20 seats minimum
Claude Enterprise, sales-assisted $20 a month, annual only, before discounts None included; all at API rates 50 seats minimum

For comparison, the individual plans are Pro at $17 a month on annual billing ($20 monthly), Max at $100 and Ultra at $200. They're for one person each, with no company admin, so they're the plans a rollout usually replaces.

How does Claude Enterprise pricing work?

An Enterprise bill has two parts: a fixed seat fee, paid annually for the seats on the plan, and usage, which is every token your team consumes in chat, Claude Code and Cowork, billed at API rates. Anthropic says there are "no per-seat usage limits and no included token allowance." Nobody runs out of an allowance, because there isn't one. The bill just grows with use.

Fig. 1What the same $20 a month buys on annual billing, as of November 2026. Source: Anthropic's pricing page and Claude Help Center billing articles.

How the usage is paid depends on how you buy:

  • Self-serve Enterprise. You buy it online. "Usage is purchased upfront in credits." Everyone draws from one balance, and "When your credits run out, usage stops" for the whole organization until an owner buys more. It's billed in US dollars only, by card or ACH bank transfer.
  • Sales-assisted Enterprise. You buy it through Anthropic's sales team, with a 50-seat minimum. Usage is "billed monthly in arrears" on an invoice, other currencies are available, and Anthropic offers "volume and committed-spend discounts", though it says "We don't publish discount rates, so figures you see elsewhere may not reflect our pricing." This is also the route for buying through AWS Marketplace.

Two terms catch buyers out. Enterprise is "only available as annual subscriptions", and on self-serve plans "Seats cannot be removed mid-term." You can add seats during the year, at a prorated charge, but you pay for the starting count until renewal. If you turn on US-only inference, usage costs 1.1 times the standard rates.

What are Claude's API rates?

Usage is priced per million tokens, and the price depends on the model: as of November 2026, Sonnet 5.5 costs $2 per million input tokens and $10 per million output tokens, Opus 5.5 costs $4 and $20, and Fable 5.2 costs $8 and $40. Haiku 5.5, the cheapest, costs $0.10 and $0.50 for prompts up to 100,000 tokens.

A token is a piece of a word. Input tokens are everything Claude reads on a request: the question, any files, and the conversation so far. Output tokens are what it writes. Anthropic's consumption guide notes that chat usage "scales with message length and conversation history", so a long conversation costs more with every reply.

To make that concrete, take one request where Claude reads 20,000 tokens (a long document and some back-and-forth) and writes 1,000. By our arithmetic from the published rates, that's about 5 cents on Sonnet 5.5, 10 cents on Opus 5.5 and 20 cents on Fable 5.2. The same work costs four times as much on the most capable model as on the everyday one. That's why model choice is the first lever Anthropic's own guide gives admins.

How much does usage add per person?

Anthropic doesn't publish a per-person figure for chat, but for coding it does: across enterprise deployments, Claude Code averages "around $13 per developer per active day and $150-250 per developer per month", with costs "below $30 per active day for 90% of users". That figure is for Claude Code alone, from Anthropic's Claude Code cost documentation.

Chat costs much less per person. Anthropic's consumption guide rates chat as "Lower intensity" and says Claude Code and Cowork "consume tokens at a significantly higher rate than standard chat". It doesn't put a dollar figure on a chat user, so you'll need your own pilot for that. What you can plan around is the shape: a few heavy users, usually developers or people running agent tasks in Cowork, will drive most of the usage bill.

Claude Team plan vs Enterprise: which costs less?

At list price, a standard Claude Team seat always costs less than an Enterprise seat for the same person, because both are $20 a month on annual billing and only the Team seat includes usage. Against a $100 premium Team seat, Enterprise costs less for anyone whose usage stays under $80 a month, and more above that.

Fig. 2Monthly cost per person on annual billing at list prices, as of November 2026. Team seats are flat up to each seat's usage limits; beyond them, usage credits are billed at API rates. Our arithmetic from Anthropic's published prices.

That break-even is simple arithmetic: an Enterprise seat costs $20 plus usage, so it matches a $100 premium seat when usage reaches $80. Two details move it in real life:

  • Team seats have limits. A standard seat gets 1.25 times the Pro plan's usage per session, and a premium seat 6.25 times, each with a weekly limit. Limits are per person and can't be shared. When someone hits theirs, they wait for the reset, or the company turns on usage credits, which are billed "at standard API pricing rates". A Team plan with usage credits on starts to behave like Enterprise for its heaviest users.
  • Team includes API credits; Enterprise doesn't. Team plans get monthly credits for building on the Claude API: $20 per standard seat and $100 per premium seat, pooled and capped at $500 a month. They don't cover chat or interactive Claude Code. Anthropic says Enterprise plans "aren't eligible."

So the price rarely decides it. The choice comes down to size and controls.

What does Claude Enterprise include that Team doesn't?

Enterprise adds the controls an IT, security or compliance team usually asks for: SCIM provisioning, audit logs, the Compliance API, role-based access, custom data retention, network-level access control, IP allowlisting, domain capture and a HIPAA-ready offering. Anthropic's comparison table marks all of these "No" for Team.

You need Plan
A company workspace for 2 to 150 people, with single sign-on and no training on your content by default Team
More headroom for a few heavy users Team, with premium seats or usage credits
More than 150 people Enterprise
SCIM, role-based access or audit logs Enterprise
A retention period you set Enterprise
Patient data and a business associate agreement Enterprise (HIPAA-ready offering)
Payment by invoice, another currency, or committed-spend discounts Enterprise, sales-assisted

Both plans include single sign-on, central billing, usage analytics and admin controls for connectors, and Anthropic says neither trains on your content by default. If you handle health data, our guide to HIPAA compliant AI covers what a business associate agreement does and doesn't cover.

How do you control Claude Enterprise costs?

With spend limits, set before anyone gets access: Enterprise lets admins cap spend for the whole organization, for groups and for individual users, and a member with no limit at any level has no cap at all. Anthropic's consumption guide says it plainly: "No limit anywhere = no limit."

Anthropic's guide recommends starting with group limits and per-user limits rather than relying on an organization-wide cap, because hitting the org limit "affects everyone simultaneously". Admins get alerts at 75% and 90% of an org-level limit. Setting a limit to unlimited doesn't make usage free: "consumption will still be billed."

The other levers are about who uses what:

  • Gate the expensive tools. Anthropic suggests rolling out Claude Code and Cowork "in waves", and only to the groups that need them, since those are where usage concentrates.
  • Decide on the unified Chat and Cowork experience. Anthropic's admin guide says the new unified experience is "off by default in existing organizations and turns on by default on November 16, 2026." Once on, it reaches every member with the User, Admin, Owner or Primary Owner role; custom roles get it only if their role includes "Chat and Cowork unified". If you want Cowork gated to some groups, an Owner can choose "Keep disabled" in Organization settings > Capabilities, or limit it by custom role. A choice an owner already made stays. Usage then shows under a "Chat and Cowork unified" line, so add that line when you total Chat and Cowork spend, or the numbers will appear to drop.
  • Set a default model. Admins can choose the model every new conversation starts on and restrict which models each role can use. Anthropic tells users that Sonnet "is your daily driver" and to keep Opus for harder work.
  • Watch the reports. Analytics shows spend by model and the top users by spend, and you can export a per-user, per-model spend report.

This is the spending-limit guardrail in our framework. Decide the limit per person and per team, who can raise it, and when you'll review it, before the first seat goes live. Our lesson on spending limits shows how to set them.

What would Claude Enterprise cost a 100-person company?

For 100 people, the seats alone cost $24,000 a year, and usage comes on top: if 20 of those people are developers using Claude Code, Anthropic's published average suggests $36,000 to $60,000 a year more for them alone. The other 80 people's chat usage is the number you have to measure.

Fig. 3An illustration at list prices, before discounts and tax. The developer range applies Anthropic's published Claude Code average of $150 to $250 per developer per month across enterprise deployments. Source: Anthropic, November 2026.

Here's how that adds up. One hundred seats at $20 a month for 12 months is $24,000. Twenty developers at Anthropic's $150 to $250 a month each is $36,000 to $60,000 a year. The chat usage for the rest of the company depends on how much they use Claude and which model they use, which is why Anthropic tells teams to "start with a small pilot group" and establish a baseline before rolling out more widely.

For comparison, the same 100 people on standard Team seats cost $24,000 a year with usage included, up to each seat's limits. Give the 20 developers premium seats instead and the bill becomes $43,200 a year: 80 standard seats at $19,200 plus 20 premium seats at $24,000. For a mixed team under 150 people that doesn't need Enterprise's controls, that's the number an Enterprise quote has to beat.

Budget from a pilot, not a price page

With a flat seat, the budget question was "how many people?" With usage billing, it's "how much work will the heaviest users hand to Claude?", and nobody knows until people start. That's the case for a two-stage proof: a small group first, with its usage and its results measured, before you sign a contract on a guess.

This applies whichever vendor you pick. ChatGPT Enterprise and Microsoft 365 Copilot also bill some usage on top of the seat, and our guide to ChatGPT Enterprise pricing goes through how OpenAI's credits work. If you're still choosing a vendor, our Claude vs Copilot comparison covers plans, data terms and admin controls.

How to budget a Claude rollout

01 · Count who already uses Claude

List who uses Claude for work today, including personal Pro, Max and Ultra plans on expense reports. The Shadow AI Audit does this in about five minutes. This gives you the seat count and a first guess at the heavy users.

02 · Decide whether you need Enterprise at all

Check the controls table above against what your IT and compliance teams need. If you're under 150 people and need none of SCIM, audit logs, custom retention or a HIPAA-ready setup, price Team first. This tells you which quote to ask for.

03 · Split people into usage groups

Put staff into four to six groups by how they'll use Claude, the range Anthropic's guide suggests: for example light chat users, heavy chat users and developers. Decide which groups get Claude Code and Cowork, and check the "Chat and Cowork unified" setting matches that decision, since it turns on by default for existing organizations from November 16, 2026. This gives you the structure for both the budget and the limits.

04 · Set limits before access

Set a monthly limit for each group and for each user, and an org-wide ceiling with alerts. On self-serve Enterprise, size the first credit purchase to the limits, not to a hope. This gives you a ceiling on the first bill.

05 · Pilot with a small group and measure

Give one or two groups access for four to six weeks. Export the spend report per user and per model, and put it next to a measure of the work, such as time saved or rework. This turns Anthropic's averages into your own numbers.

06 · Size the contract on what you saw

Choose Team or Enterprise, the seat count and the usage budget from the pilot, with room for growth. If you're near 50 seats or above, ask Anthropic's sales team about committed-spend discounts. Put a review date in the calendar a month before renewal. This makes the second year's number one you can defend.

Moving from Team to Enterprise keeps your chats, projects and members, but Anthropic says the self-serve migration "is not reversible", and some features that are on by default in Team, such as skills and code execution, start switched off in Enterprise. Plan the switch; don't click through it.

If you'd like help with the numbers, our AI implementation work starts with the list of what's in use and the pilot, and it's part of deploying AI without a bill nobody can explain.

What's still open

  • Discounts. Anthropic offers volume and committed-spend discounts on sales-assisted Enterprise plans but doesn't publish them. Treat any discount figure you read online as unconfirmed.
  • Model prices. API rates change as Anthropic releases and retires models, and the Enterprise bill follows them. Check the rate for the model your company defaults to.
  • Legacy Enterprise plans. Some older Enterprise customers still have seat-based plans with usage allowances. Anthropic says they move to the usage-based seat at their next renewal, so expect the bill's shape to change then.

We'll update this article when Anthropic changes its seat prices, minimums or how usage is billed. The date at the top of the page shows when we last checked.

Sources

Questions

Common questions

Start with step one

Find out who in your company already pays for Claude.

Every Claude quote starts with a seat count and a guess at who the heavy users are. The free Shadow AI Audit lists the AI already in use, including personal Claude Pro and Max plans on expense reports, in about five minutes.