Claude Enterprise pricing: $20 a seat plus usage, Claude Team cost per seat, and the API rates Anthropic bills on top.
We sell a competing product, so read this with that in mind. Every Anthropic figure below was read off claude.com/pricing and the Anthropic model pricing documentation on August 30, 2026, with the Team and Enterprise tab open and the page served to a United States address. That last detail matters: two vendors we checked this month priced in local currency rather than dollars when loaded from Europe, so a lot of the numbers circulating in roundups are conversions of somebody else's regional rate. Nothing here is estimated, and where Anthropic publishes nothing we say so instead of guessing.
WorkAgent console
Delegate a task · watch it run · get the result
Pick a task to delegate
It works the task end to end and hands back .
| Company | Contact | Fit | |
|---|---|---|---|
Subject:
Research, lead gen, outreach, inbox and reporting, delegated and done.
Like what it did? Put WorkAgent on your real work.
In short
Last updated August 2026
Claude Enterprise costs $20 per seat per month plus usage billed at Anthropic API rates, which is the same seat price as a Claude Team standard seat. That makes the two $20 figures mean opposite things: on Team, $20 is the whole bill for that seat, and on Enterprise, $20 is the floor before a single token is counted. Anthropic is the only major chat assistant vendor that publishes an Enterprise seat price at all, because OpenAI lists ChatGPT Enterprise as custom pricing with no rate anywhere. Claude Team is $20 a seat billed annually or $25 billed monthly for a Standard seat, and $100 or $125 for a Premium seat with five times the usage, for teams of 2 to 150 people.
- Claude Enterprise
- $20 per seat, per month, plus usage
- Claude Team, Standard seat
- $20 per seat, per month, billed annually
- Claude Team, Premium seat
- $100 per seat, per month, billed annually
- WorkAgent
- $149 / month, flat
Calculator
Claude Enterprise pricing calculator: model the seat fee and the usage meter together
Set your seat count, pick a model and estimate how hard your team works it. The arithmetic runs in your browser at Anthropic published rates, and it shows you the Claude Team seats side by side so you can see which plan is actually cheaper for your shape. No email, no form, no sales call.
Your deployment
$20 a seat on Enterprise, before any usage.
: per million in, per million out.
21 working days a month. Agentic and batch work runs far higher than typing does.
Prices 80 percent of input at the cache hit rate, which is exactly one tenth of base input on every Claude model.
What Claude Enterprise costs you
Per month, all in
seats, blended each
Twelve months
Seat fee plus metered usage
The two halves of the Enterprise bill
- Seat fee, at $20
- Usage at API rates, exchanges
- Usage per seat
The same headcount on the capped plans
- Claude Team, Standard seats, all in
- Claude Team, Premium seats, all in
- WorkAgent, flat
- $149.00
Team seats are capped rather than metered, so they cannot overrun. They also cannot exceed the limit when you need them to.
At this workload each Enterprise seat is consuming of usage a month, which is past the $80 of headroom between a $20 Enterprise seat and a $100 Team Premium seat. On price alone the capped Premium seat is now the cheaper option for these users, so buy Enterprise here for the governance and the absence of a limit, not to save money.
Above 150 people Claude Team is no longer the plan you can buy. Anthropic scopes Team to 2 to 150 and moves everyone larger to Enterprise. Unlike OpenAI, which publishes no ChatGPT Enterprise rate at all, Anthropic does publish the seat price on the other side of that line.
Rates read off claude.com/pricing and the Anthropic model pricing documentation on August 30, 2026, served to a United States address: Claude Enterprise $20 a seat plus usage at API rates, Claude Team Standard $20 annually and $25 monthly, Premium $100 annually and $125 monthly. Token rates per million: Opus 5 $5 input and $25 output, Sonnet 5 $2 and $10, Haiku 4.5 $1 and $5. Cache hits are one tenth of the base input rate. The caching toggle prices 80 percent of your input tokens at the cache hit rate, which is a reasonable approximation of a repeated long context and not a guarantee. Working month is 21 days. Prices exclude tax and Anthropic can change them.
What it handles
Which Claude plan to buy for which job, and where it is the wrong tool
What Anthropic charges per seat, why the Enterprise seat price is a floor rather than a price, the 150 person ceiling on Team, and where a flat rate agent is the better buy.
Giving a whole department a governed AI chat window
Claude Team, in most cases. It is the assistant your staff already know wrapped in SSO, central billing, admin control over connectors, enterprise deployment of the desktop app and a default of not training on your content. If the requirement is that 40 people each get their own window, buy a per seat product and stop reading price comparisons.
Regulated work, audit trails and access control
Claude Enterprise. This is where SCIM, audit logs, role based access with fine grained permissioning, a compliance API, custom data retention, network level access control, IP allowlisting and a HIPAA ready offering live. None of those appear on Team, and Anthropic charges the same $20 seat for them.
Heavy users who keep running into limits
Claude Enterprise, because it swaps the limit for a meter. A Team Premium seat gives five times a standard seat and then stops. Enterprise bills usage at API rates, so a power user produces a larger invoice rather than a wall. Decide which of those two failure modes your finance team prefers.
Software teams that also want a coding assistant
Either Claude plan, and this is a genuine cost advantage. Claude Code is inside the seat rather than sold separately. Microsoft charges for GitHub Copilot on top of Microsoft 365 Copilot at $10, $39 or $100 a head, so 20 developers is up to $2,000 a month of difference that seat comparisons routinely leave out.
Getting a specific job finished without anyone operating a tool
WorkAgent. "Research these 40 firms, find who runs operations at each, draft a first touch and put the list somewhere I can check it" is a briefed outcome, not a chat window. Nobody has to log in, prompt it, or learn what to ask, which is the part per seat pricing quietly assumes they will do.
A bill that does not move when headcount does
WorkAgent. $149 flat covers the account. Claude Team and Claude Enterprise are both priced per person, so every hire, contractor and seasonal helper adds $20 or $100 a month whether or not they open it that week, and Enterprise adds their tokens on top.
Why it works
Six things about Claude Enterprise pricing that comparisons get wrong
The same $20 means two opposite things
A Claude Team standard seat is $20 and that is the entire bill for that seat. A Claude Enterprise seat is also $20 and that is the floor, because Anthropic prices Enterprise as "Seat price + usage at API rates" and notes that usage cost scales with model and task. One number is a ceiling, the other is a starting point. Any comparison that puts them side by side as equal prices is comparing two different kinds of number.
Anthropic publishes an Enterprise price at all, which is unusual
OpenAI lists ChatGPT Enterprise as custom pricing with a contact form and no rate, no minimum and no volume table. Google publishes seat prices for Gemini Enterprise. Glean publishes a full rate card and no price. Sierra publishes nothing. Anthropic sits at the transparent end: a published seat figure plus a published, per model token rate you can compute against before you ever speak to sales.
Enterprise governance costs nothing per seat
The usual industry pattern is that SCIM, audit logs and role based access are the upsell that justifies a higher seat price. Anthropic does not charge for them in the seat. SCIM, audit logs, RBAC, the compliance API, IP allowlisting and the HIPAA ready offering all arrive at the same $20. The premium is moved entirely into metered usage, which is a different bet: cheap for light governed users, open ended for heavy ones.
Cache hits are exactly one tenth of the base input rate, on every model
Opus 5 is $5 per million input tokens and $0.50 on cache hits. Sonnet 5 is $2 and $0.20. Haiku 4.5 is $1 and $0.10. Fable 5 is $10 and $1. The ratio is exactly ten to one across the whole range, so on Enterprise, where you actually pay for input, prompt caching a long standing context is the single largest lever on the usage half of your bill.
A newer tokenizer counts about 30 percent more tokens
Anthropic documents that Claude 4.7 and later models use a newer tokenizer producing approximately 30 percent more tokens for the same text. That means a per token rate is not comparable across model generations. Opus 4.1 at $15 per million input against Opus 5 at $5 looks like a three times cut on the rate card, but priced per page of actual text it is closer to 2.3 times. Real, large, and smaller than the headline.
The Team ceiling is a headcount, not a missing feature
Anthropic scopes Claude Team to teams of 2 to 150 people. OpenAI scopes ChatGPT Business to 2 to 200. So the same growing company hits Anthropic's ceiling 50 employees earlier. The difference is what waits on the other side: crossing Anthropic's line lands you on a plan with a published $20 seat, and crossing OpenAI's line lands you on a plan with no published price at all.
Compare
Claude Team, Claude Enterprise and WorkAgent, side by side
Anthropic columns read off claude.com/pricing with the Team and Enterprise tab open, served to a United States address, on August 30, 2026. Prices change, so confirm with Anthropic before you buy.
| Claude Team, Standard | Claude Team, Premium | Claude Enterprise | WorkAgent | |
|---|---|---|---|---|
| Published seat price, billed annually | $20 per seat, per month | $100 per seat, per month | $20 per seat, per month | $149 a month, flat |
| Published seat price, billed monthly | $25 per seat, per month | $125 per seat, per month | Seat plus usage, no monthly uplift published | $149 a month, flat |
| Is the seat price the whole bill | Yes | Yes | No, usage is billed at API rates on top | Yes |
| What the price buys | One named user | One named user | One named user plus their tokens | The whole account |
| Ten people | $200 a month | $1,000 a month | $200 a month plus usage | $149 a month |
| Fifty people | $1,000 a month | $5,000 a month | $1,000 a month plus usage | $149 a month |
| Team size allowed | 2 to 150 people | 2 to 150 people | No published ceiling | Seats are not counted |
| Usage | More than Pro, limits apply | Five times a Standard seat | Metered, no limit to hit | Included in the flat price |
| Mix seat types in one workspace | Yes | Yes | Not applicable | Not applicable |
| Single sign on | Yes | Yes | Yes | Not seat based |
| SCIM | No | No | Yes | Not seat based |
| Audit logs | No | No | Yes | Not seat based |
| Role based access, fine grained | No | No | Yes | Not seat based |
| Compliance API | No | No | Yes | No |
| Custom data retention controls | No | No | Yes | No |
| IP allowlisting and network access control | No | No | Yes | No |
| HIPAA ready offering | No | No | Available | No |
| Admin spend limits per user and per org | No | No | Yes | Not applicable |
| Claude Code included in the seat | Yes | Yes | Yes | Not applicable |
| Training on your content by default | No | No | No | No |
| Buy without talking to sales | Yes | Yes | Yes, self serve Enterprise exists | Yes |
How much does Claude Enterprise cost?
Twenty dollars per seat per month, plus usage billed at Anthropic API rates. Anthropic states it on the pricing page in exactly those terms: the Enterprise seat header reads "Seat price + usage at API rates", the figure beneath it is $20/seat, and the qualifier is that usage cost scales with model and task. That is the complete published answer, and it is more than most vendors in this category give you.
The part that trips buyers up is that the same $20 also buys a Claude Team standard seat, where it is the entire bill. So the honest framing is that Claude Enterprise is never cheaper than Claude Team for an identical seat. It is the same seat fee with a meter attached and a governance stack switched on. You move to Enterprise for SCIM, audit logs, role based access, the compliance API, custom retention, IP allowlisting and the HIPAA ready offering, or because your heaviest people keep hitting Team limits, not to save money.
What you cannot do is put a single monthly number on Enterprise without modelling the usage half yourself, which is why the calculator above exists. Anthropic gives you everything you need to do that arithmetic: a published seat fee and a published per million token rate for every model. Compare that with ChatGPT Enterprise pricing, where OpenAI publishes the word "custom" and nothing else.
What Claude Team costs per user, and the standard against premium seat split
Claude Team sells two seat types and most write ups mention one. A Standard seat is $20 per user per month billed annually, or $25 billed monthly. A Premium seat, which Anthropic describes as five times more usage than a standard seat, is $100 billed annually or $125 billed monthly. Anthropic explicitly supports mixing the two in a single workspace, so a real invoice usually sits somewhere between the two rates rather than on either one.
The monthly penalty is exactly 25 percent on both seat types, which is worth stating precisely because it is a bigger gap than the individual plans carry. Claude Pro is $17 a month on the annual subscription, billed as $200 up front, against $20 billed monthly. That is a 20 percent uplift for individuals against 25 percent for teams. Committing for a year is worth more to a team buyer than to a single user, which is the opposite of what most people assume.
Team is scoped to 2 to 150 people. Below two, you are on Pro or Max as an individual. Above 150, Anthropic moves you to Enterprise. Because Enterprise publishes a seat rate, crossing that line is a budgeting exercise rather than a procurement negotiation, which is genuinely unusual in this market.
How the usage half is priced, model by model
Enterprise usage is billed at the standard Anthropic API rates, which are published per million tokens and split into input and output. On the current line up: Claude Opus 5 is $5 per million input tokens and $25 per million output, Claude Sonnet 5 is $2 and $10, and Claude Haiku 4.5 is $1 and $5. Claude Fable 5, the largest model, is $10 and $50. Retired Opus 4.1 and Opus 4 were $15 and $75, which is where a lot of stale cost estimates still anchor.
Prompt caching is the lever that matters most. Cache hits and refreshes are priced at exactly one tenth of the base input rate on every model in the range: $0.50 on Opus 5, $0.20 on Sonnet 5, $0.10 on Haiku 4.5, $1 on Fable 5. Cache writes cost 1.25 times base for a five minute cache and twice base for a one hour cache. If your team repeatedly sends the same long context, a policy handbook, a codebase, a contract set, caching moves the dominant line item down by an order of magnitude.
One dated detail worth capturing, because it changed recently and a lot of published estimates have not caught up. Sonnet 5 launched at $2 and $10 as introductory pricing due to expire on August 31, 2026, with a scheduled increase to $3 and $15 on September 1. Anthropic documentation now states that the introductory rate is the standard price and that the increase will not occur. If you budgeted a 50 percent Sonnet rise for next month, remove it.
The tokenizer detail that breaks per token comparisons
Anthropic documents that Claude 4.7 and later models, along with Claude Mythos Preview, use a newer tokenizer that produces approximately 30 percent more tokens for the same text, with the exact increase depending on content and workload shape. Claude Sonnet 4.6 and earlier use the previous tokenizer. This is the kind of thing that never appears in a pricing comparison and quietly changes the answer.
Work the arithmetic. Opus 4.1 charged $15 per million input tokens; Opus 5 charges $5. On the rate card that is a three times reduction. Priced per page of your actual text, the newer model turns the same document into roughly 1.3 million tokens where the old one made 1 million, so the real comparison is $15 against $6.50. That is a 2.3 times cut. Still a large reduction, and meaningfully smaller than the headline number implies.
The general rule this produces is worth keeping: a published per token rate is only comparable between two models that tokenize the same way. Across vendors it is worse, because nobody publishes a conversion. If you are comparing Claude usage costs against another provider per token, you are comparing units that do not match, and the only sound method is to price the same real workload through both.
When Claude Enterprise usage overtakes a Claude Team premium seat
This is the calculation that actually decides the plan, and it is easy to run once the rates are in front of you. A Team Premium seat costs $100 all in. An Enterprise seat costs $20, leaving $80 of headroom before Enterprise becomes the more expensive choice for that person. So the question is how much usage $80 buys at API rates.
Take a working assumption of 4,000 input and 800 output tokens per exchange on Sonnet 5, with no caching. That is $0.008 of input and $0.008 of output, about $0.016 an exchange. Eighty dollars buys roughly 5,000 exchanges a month, or around 240 every working day for one person. Turn caching on for the repeated part of the context and the input side drops tenfold, pushing the crossover well past what a human can generate by typing.
The practical conclusion is that for ordinary chat use, Enterprise usage rarely approaches $80 a head and the governance arrives close to free. Where it inverts is automated and agentic work: long document processing, code agents chewing through repositories, batch jobs. Those consume tokens at machine speed rather than typing speed, and that is exactly where a metered plan stops being cheaper than a capped one. Model your heaviest workload, not your average one.
Where a flat rate agent is the better purchase
Per seat pricing assumes the bottleneck is access, that if everyone has a window onto a good model the work gets done. Often that is right, and when it is, buy Claude Team and stop. But a lot of the work people actually want automated is not conversational. Nobody wants to prompt their way through building a prospect list every Monday, they want the list.
WorkAgent is priced the other way round: $149 a month, flat, for the account, with no seat count and no token meter. You brief an outcome, it does the research, the drafting, the checking and the delivery, and the price does not move when you hire someone or when a job turns out to need more thinking than you expected. For a five person team the arithmetic is not close in either direction, so pick on shape rather than price. For a growing operations team where three people would need the tool but only occasionally, flat rate usually wins.
The honest comparison across the pricing cluster is worth reading before you commit to any of these: ChatGPT Enterprise pricing for OpenAI published rates, Gemini Enterprise pricing for the Google seat ladder, Copilot Studio pricing for Microsoft consumption model, and AI agent pricing for how the per seat, per credit and flat rate models compare as models.
FAQ
Questions buyers ask about Claude Enterprise pricing
How does Claude Enterprise pricing work?
Claude Enterprise is priced as a seat fee plus metered usage. Anthropic charges $20 per seat per month and then bills the tokens your people consume at standard API rates, which vary by model and task. The seat fee buys the governance stack, SCIM, audit logs, role based access, compliance API, custom retention and IP allowlisting. The usage half you model yourself.
How much does Claude Enterprise cost?
Twenty dollars per seat per month, plus usage at API rates. That is the figure Anthropic publishes on claude.com/pricing under the heading "Seat price + usage at API rates". There is no published minimum seat count and no published volume table, so the seat half is fixed and the usage half depends entirely on how heavily your team works the models.
What is Claude Enterprise pricing per user?
The published per user figure is $20 a month. Unlike Claude Team, that is not the full cost of the user, because their token consumption is billed on top at API rates. For a typical chat user on Sonnet 5, usage tends to land well under the $80 that would make an Enterprise seat cost more than a $100 Team Premium seat.
Claude Enterprise vs Team: which should we buy?
Buy Team if you want a predictable, capped bill and do not need SCIM, audit logs, role based access or a HIPAA ready offering. Buy Enterprise if compliance requires those controls, if you exceed 150 people, or if heavy users keep hitting Team limits and you would rather receive a larger invoice than a wall. Enterprise is never cheaper per seat.
Is there a minimum number of seats for Claude Enterprise?
Anthropic does not publish a seat minimum for Claude Enterprise. The pricing page offers both a self serve Enterprise plan you can start today without contacting sales and a sales assisted Enterprise plan for tailored terms, which suggests small Enterprise deployments are supported. Claude Team, by contrast, does state a range: 2 to 150 people.
What is the difference between a Claude Team standard and premium seat?
Usage, and nothing else in the published feature list. A Standard seat is $20 a month billed annually and $25 monthly. A Premium seat is $100 annually and $125 monthly, and Anthropic describes it as five times more usage than a standard seat. You can mix both types in one workspace, so heavy users can be upgraded individually rather than upgrading the whole team.
How much is Claude Team per user?
Twenty dollars per user per month billed annually, or $25 billed monthly, for a Standard seat. A Premium seat with five times the usage is $100 annually or $125 monthly. The monthly penalty is exactly 25 percent on both. Claude Team covers teams of 2 to 150 people and includes Claude Code, single sign on, central billing and no model training on your content by default.
Does Claude Enterprise charge per token?
Yes, on top of the seat. Anthropic bills Enterprise usage at the same published API rates it charges developers, so the meter is per million tokens and splits into input and output. This is different from Claude Team, where the seat price is the whole bill and usage is governed by limits rather than by a meter.
How much does Claude cost per million tokens?
On current models: Claude Opus 5 is $5 per million input tokens and $25 per million output. Claude Sonnet 5 is $2 and $10. Claude Haiku 4.5 is $1 and $5. Claude Fable 5 is $10 and $50. Cache hits and refreshes are priced at exactly one tenth of the base input rate on every model, so caching repeated context is the largest available saving.
Is Claude Sonnet 5 pricing going up in September 2026?
No. Sonnet 5 launched at $2 per million input and $10 per million output as introductory pricing due to end on August 31, 2026, with an increase to $3 and $15 scheduled for September 1. Anthropic documentation now states that the introductory rate has become the standard price and the scheduled increase will not occur.
Is Claude Enterprise HIPAA compliant?
Anthropic lists a HIPAA ready offering as available on Claude Enterprise. It is not present on Claude Team or on the individual plans. As with every HIPAA arrangement, availability of the offering is not the same as being covered, so you need the business associate agreement executed before any protected health information goes near the product.
Is Claude Code included in a Claude Team or Enterprise seat?
Yes. Claude Code, Claude Cowork, Claude Design and Claude Science are all listed as included in the Claude Team seat, and Enterprise carries all Team features. This is a real cost difference against Microsoft, which sells GitHub Copilot separately from Microsoft 365 Copilot at $10, $39 or $100 per developer per month.
Claude Enterprise vs ChatGPT Enterprise pricing: which is cheaper?
You cannot compare them directly, because OpenAI publishes no price for ChatGPT Enterprise at all, only the words "custom pricing" and a sales contact. Anthropic publishes $20 per seat plus usage. The tier beneath each is directly comparable and identical: Claude Team and ChatGPT Business both charge $20 and $25 for a standard seat and $100 and $125 for a premium one.
How much is Claude Enterprise for 5 users?
Five Enterprise seats is $100 a month in seat fees, plus whatever those five people consume in tokens. At a working assumption of 4,000 input and 800 output tokens per exchange on Sonnet 5 with no caching, each exchange is roughly $0.016, so a busy user doing 100 exchanges a working day adds about $34 a month. Five such users lands near $270 all in.
What are the Claude Enterprise usage limits?
Enterprise replaces limits with a meter, which is the main behavioural difference from Team. Anthropic publishes usage limits as a separate document rather than as message counts on the pricing page, and describes the shape rather than the number: Pro gets more than Free, a Team Standard seat gets more than Pro, and a Team Premium seat gets five times a Standard seat.
Does Anthropic publish nonprofit or education pricing for Claude?
Not as a rate. Anthropic lists a Nonprofits solutions page and a separate Education plan offering university wide access for students, faculty and staff at what it calls discounted rates, but neither carries a published discount percentage or seat price on the pricing page. If you qualify for either, the number has to come from Anthropic rather than from a public page.
Put it to work
See what a flat rate agent does on each job, with one price that carries no seat count and no token meter:
AI research agent that does your market and prospect research.
Market, competitor and prospect research, summarized.
ExploreAI lead generation agent that builds your lead list.
Find, enrich and qualify prospects into a clean list.
ExploreAI outreach agent that writes and sends your follow-ups.
Personalized cold emails and follow-up sequences.
ExploreAI inbox agent that triages and drafts your replies.
Triage the inbox and draft the replies.
ExploreAI CRM assistant and AI CRM agent that handles CRM updates and keeps records clean.
Log activity and keep records clean, automatically.
ExploreAI data entry agent that does the data entry for you.
Move, format and clean data between tools.
ExploreCompare the approaches: see AI virtual assistant, AI virtual agent, AI assistant for business, AI personal assistant for business, AI agent for small business, AI assistant for small business, AI SDR, AI executive assistant, AI agent software, AI for real estate agents, AI for real estate investors, AI operations assistant, AI marketing assistant, AI lead generation software, AI agents for solopreneurs, AI sales assistant software, AI agents for agencies, AI agent for consultants, AI agent for insurance agents, AI for property managers, AI agents for financial advisors, AI for mortgage brokers, AI agent for accounting firms, AI for law firms, AI for roofing companies, AI for construction companies, AI agent pricing and AI employee, compare us against Lindy AI and Relevance AI, Artisan AI, Zapier Agents, Make.com, Gumloop, Apollo.io and Clay, check GoHighLevel AI Employee pricing and Agentforce pricing, run the numbers in the Agentforce pricing calculator and Copilot Studio pricing and Zapier Agents pricing and Clay pricing and Artisan AI pricing and Gemini Enterprise pricing and Glean pricing and Agentforce vs Copilot Studio, or read how the AI agent works and AI worker pricing.
Make your first AI hire.
Try the demo above, then join the queue to put WorkAgent on your real tasks. Plans from $149/mo.