SVENNISAIClaude pricing in the UK, and which plan you need
Anthropic prices its UK seats in pounds, from its own UK list. An agent's usage is paid in US dollars, on use, and everything is billed to your company's account.
Book a 30-minute call · with a Claude Certified Architect from our team.
Source: Anthropic, Claude pricing (the UK price list, in GBP), the list the page shows a visitor in the UK.
Four ways of working with Claude
Three are for people and the fourth is for software. The API is for a company that wants a process to run on its own.
- For peopleChatEveryone in the company. You ask, draft, analyse a document or a spreadsheet. This is where everyone starts.
- For peopleCoworkOffice roles: sales, finance, operations, HR. You hand it a whole piece of work, not a question: it works from your files, browser and connected apps and brings back the result. It is on every paid plan today. Anthropic is now merging it into Claude, on Pro and Max first, so the name may go while the function stays.
- For peopleClaude CodeDevelopers and system administrators. Claude for developers, inside their own tools; our team builds with it. Finance has no need of it, but it comes with the seat.
- For softwareAPIA company that wants a process to run on its own. Software that calls Claude with no person at the keyboard: an assistant that answers customers, a program that reads incoming invoices. We build it, or your own team does.
The four plans and what each costs
Anthropic's UK list, in pounds. You buy the plan from Anthropic, on your company's account.
Anthropic's own UK page shows the same amounts with VAT at 20% added. There is also a Free plan to try Claude: chat on the web, desktop and phone, without Claude Code. Seat minimums, discounts and contract terms are Anthropic's.
On every paid plan; it draws on the same usage limit as the rest of the plan.
On the desktop (macOS and Windows), on every paid plan, today. On the web and on phones, in beta on Pro, Max and Team; on Enterprise when an admin turns it on. Cloud sessions (on Anthropic's infrastructure, not the employee's computer): in beta on Team and Enterprise.
SSO (sign-in with the company account), central billing and domain verification. Spending limits for extra usage, not for the seat price. Search connectors for Google Drive, Gmail, Microsoft 365, Slack and others.
SCIM, audit logs, the Compliance API (records for audit). Custom data retention, roles and permissions (RBAC), IP allowlisting. Domain capture (accounts on the company's domain come under its administration). The HIPAA offer, a US health-sector requirement that does not concern a UK firm.
On Team and Enterprise, off by default. On Free, Pro and Max you turn it off yourself in the privacy settings (opt-out).
Last reviewed: 20 October 2026 for the seat list; 13 October 2026 for what the plans include.
If this is you, start here
A seat is one person's licence on Claude Team or Claude Enterprise, bought directly from Anthropic. The table shows the plan to start on; you can move to another later.
| Your situation | Start here | Why |
|---|---|---|
| One person trying Claude | Pro | The individual subscription: see what Claude does on your own work, with no commitment for the company. |
| A team of 2 to 150 people working with company data | Team | Never Pro or Max for client data: they are personal plans on consumer terms, without the data processing agreement UK GDPR expects from a processor. Team runs on Anthropic's commercial terms, with central administration, and Anthropic does not train on your content by default. |
| Audit logs, SCIM (accounts opened and closed automatically from the company directory), custom data retention, IP allowlisting (access only from the company's networks); or more than 150 people | Enterprise | Those controls exist on this plan only. A minimum of 20 seats and an annual commitment, according to Anthropic's page. On top of the seat, every token is paid at API rates, so the total depends on use. |
| Software that must answer customers or process documents on its own | The API, on your own account | You pay for the text processed, counted in tokens, not for subscriptions per person. We build and run it on your own Anthropic or AWS account. |
| A team of developers | Claude Code, included in Team or Enterprise seats | It is not a separate licence: every paid seat has Claude Code, and it draws on the same usage limit. |
Claude Pilot B sets up Claude Team on your company's account. The base price covers the company account for the whole team, the existing connectors, three real tasks, the company's AI policy and two Claude ambassadors.
API prices for the models we build with
A token is a piece of a word, the unit Anthropic counts text in. As a guide, an ordinary email is a few hundred tokens and a page of a document about a thousand. Prices are per million tokens, for input (what you send) and output (what comes back).
| Model | Input | Output | For |
|---|---|---|---|
| Claude Haiku 5.5 | 0.10 USD for requests of up to 100,000 input tokens; 0.50 USD above | 0.50 USD for the same requests; 2.50 USD above 100,000 input tokens. The output price also depends on the size of the input | High volume and fast replies: classification, routing, live chat with customers. |
| Claude Sonnet 5.5 | 2 USD | 10 USD | Clearly defined tasks and documents ready to send. |
| Claude Opus 5.5 | 4 USD | 20 USD | The everyday model for complex work. |
| Claude Fable 5.1 | 10 USD | 50 USD | Long pieces of work with many stages, and the hardest problems. Anthropic keeps requests for 30 days, so it does not run with zero data retention; zero retention exists on some models, not on Fable 5.1. |
Prices in USD per million tokens, checked on 13 October 2026; batch, with the answer later, costs half. Cached input is the repeated part of a request. It costs 5% of input on Opus and Sonnet 5.5, 2.5% on Fable 5.1 and 10% on Haiku. Anthropic picks the processing region by default at the standard price; US-only processing, on request, costs 1.1 times as much.
Source: platform.claude.com, the pricing page. The figures are Anthropic's and may change. Last reviewed: 13 October 2026.
What an agent costs to run, and where it runs
The agents we build run on your own account, by default on Amazon Bedrock in the London region. The rates below are AWS's for that region, in USD, read on 9 October 2026. At a few thousand cases a month an agent costs tens of pounds; at 10,000 cases it passes a hundred.
| Model | Input, USD per million tokens | Output, USD per million tokens | Where it runs |
|---|---|---|---|
| Claude Sonnet 5.5 | 2.2 USD | 11 USD | From London through EU routing: processed within Europe, not always in the UK |
| Claude Haiku 5.5 | 0.11 USD | 0.55 USD | From London through EU routing: processed within Europe, not always in the UK |
| Claude Opus 5.5 | 4 USD | 20 USD | In-region in London (eu-west-2): inference runs in the UK |
An email or a ticket, with its thread and the instructions, is about 4,000 input tokens. A drafted reply or a filled record is about 800 output tokens. AWS or Anthropic bills the usage in US dollars, on use, with a budget set from the start.
AWS bills Bedrock in US dollars. The calculator converts at the Bank of England spot rate XUDLUSS of 8 October 2026: 1.3226 US dollars to the pound. Your bank's rate applies on the day. Sources: AWS Price List API, Amazon Bedrock, eu-west-2 (published 8 October 2026); AWS, model support by Region, read 10 October 2026; Bedrock prices read 9 October 2026.
What it costs, in figures
Four sums with the formula shown: Team seats from the UK list in pounds, Enterprise and the API in USD. Put your own numbers in their place.
| The sum | Formula | Result | Worth knowing |
|---|---|---|---|
| Ten people on Team standard, billed annually | 10 seats × 15 GBP × 12 months | 1,800 GBP a year | Premium only for whoever often reaches the usage limit. |
| Seven standard and three premium, billed annually | 7 × 15 GBP × 12 months + 3 × 75 GBP × 12 months | 3,960 GBP a year | The three heavy users get five times the usage of the others. |
| Forty people with audit logs, on Enterprise | 40 seats × 20 USD × 12 months | 9,600 USD a year, for the seats alone | Usage is paid on top, for every token, at API rates; Anthropic estimates it with you. |
| A customer assistant on the API, 3,000 conversations a month | 3,000 conversations × 4 question and answer turns = 12,000; × 3,000 input tokens = 36 million; × 300 output tokens = 3.6 million; no cache. Claude Haiku 5.5: 36 × 0.10 USD + 3.6 × 0.50 USD. Claude Sonnet 5.5: 36 × 2 USD + 3.6 × 10 USD | about 5.40 USD a month on Claude Haiku 5.5; 108 USD a month on Claude Sonnet 5.5 | An ordinary assistant combines the two and comes to tens of dollars a month; that is where the line on our services page comes from. |
Your seats and your agents, a month, in pounds
Start from a firm like yours, then change any figure. Seats at Anthropic's UK list; agent usage at Bedrock London rates.
Your estimate, a month
Prices exclude VAT. Seats are Anthropic's UK list prices in pounds, read 20 October 2026. AWS bills agent usage in USD, converted here at the Bank of England spot rate XUDLUSS for 8 October 2026. Your bank's rate applies on the day. Bedrock prices read on 9 October 2026.
Pro and Max are personal plans with different data terms; we recommend Team for business use.
Our own work is priced separately, in the written offer: see the published prices.
Asked about Claude's prices in the UK
Does Anthropic bill UK customers in pounds?
Yes, for the plans bought online: Anthropic's UK list for Pro, Max and Team is in pounds. This page carries that list, read on the date shown. Enterprise is contracted in US dollars.
Can I use Pro for client data?
We advise against it. Pro and Max are personal plans on consumer terms, without the data processing agreement UK GDPR expects from a processor, and training is opt-out. For company or client data, from two people upwards, start on Team.
Team or Enterprise for 12 people?
Team, in almost every case. Twelve standard seats billed annually are 12 × 15 GBP × 12 months = 2,160 GBP a year; billed monthly, 12 × 18.75 GBP = 225 GBP a month. Enterprise earns its place only for audit logs, SCIM, custom retention or IP allowlisting, from 20 seats, with usage paid on top.
What is the usage limit, and what happens when I reach it?
Anthropic sets the limit per five-hour session and per week, not in messages, so it depends on how long the work is. Team standard has 1.25 times the per-session usage of Pro and premium 6.25 times; Claude Code draws on the same limit. At the limit you wait for the reset or, on Team, buy extra usage within your spending limits.
What is Cowork?
Cowork is what Claude does when you give it a task, not a question. It reads the folders and apps you allowed, does the steps, shows you each one and stops where you told it to. It comes with every paid Claude plan and is not a separate product. Claude Cowork: what it does and how you start
Where does the data on a seat go? Is there a data processing agreement?
Anthropic stores Team and Enterprise data in the US and may process it in the US, Europe, Asia or Australia. No seat offers UK or EU residency today. Team and Enterprise run on Anthropic's commercial terms with a data processing agreement; Pro and Max are on personal terms. The agents we build run in your own AWS account from London, which is a separate decision made in the offer. Data and security: the layers, the policy, the law
What does an agent's usage cost in pounds?
It depends on the model and the volume. On Claude Sonnet 5.5 from London, an agent costs about 7 GBP a month at 500 cases. At 10,000 cases it is about 133 GBP, at the Bank of England rate of 8 October 2026. We work out the exact figure in the Claude Readiness Workshop, on your process and your volume.
Can you buy the licences for us?
No, and that is better for your company: the Anthropic account is yours, with your data and your keys. We set up the organisation, the seats, SSO and the connectors and run the agents on that account; Anthropic invoices you directly. If you stop working with us, the account, the data and the history stay with your company.
What happens to the seat of an employee who leaves?
The answer is on Anthropic's billing page in the company account; we do not broker seats. The same page shows how annual billing works on Team.
What would 12 office staff do with Claude on an ordinary day?
The dispatcher can give it the orders and the drivers and get the day's schedule and the messages to customers, to check first. The bookkeeper can give it the bank statement and the invoices and get them matched, with a list of what is unclear. The customer service colleague gets a drafted reply to every email, with the data from the connected system.
What happens when Anthropic changes its prices?
We read the list again and change the figures and the date on this page. Every figure names its source, and the date it was read sits beside it. Anything we cannot check comes off the page until we have checked it.
Who wrote this page, and what does your work cost?
Svennis Cloud Solutions Ltd, registered in England and Wales, with its office in Manchester. We do not sell licences. The prices of our own work, from the Claude Readiness Workshop to Managed AI Operations, are below and on the services page. Our services and prices
Where the figures come from: Anthropic, Claude pricing (the UK price list, in GBP) (the list behind it: Anthropic's price service for the UK), read 20 October 2026. Prices are Anthropic's and may change; Claude is a trademark of Anthropic, PBC.
What our work costs, published in full
Every step has a published price and a decision at its end. The Workshop is credited in full against a pilot signed within 60 days.
- Step 1Claude Readiness Workshop2 weeks: the processes worth doing, with the value and the cost of each, and the targets for the pilot.2,500 GBP
- Step 2, option AClaude Pilot A: an AI agent on Claude30 days. The base price covers one process end to end, one channel, the existing connectors, one new MCP server and the skills.from 16,800 GBP
- Step 2, option BClaude Pilot B: Claude Team for your team30 days. The base price covers the company account for the whole team, the existing connectors, three real tasks, the company's AI policy and two Claude ambassadors.from 12,500 GBP
- Step 3Department ActivationA 2-week sprint per department, with 5 to 8 tested skills.from 5,600 GBP per department
- Step 4Managed AI OperationsMonthly, no minimum term: Lite, Standard from 1,650 GBP a month, or Full, priced on scope.from 550 GBP a month
Every price we charge. All prices are in GBP and exclude VAT.
Bring your headcount. Leave with the real figure.
In 30 minutes the estimate becomes a plan: how many seats, which agent is worth doing first and what it costs at the published prices. You speak to a Claude Certified Architect from our team, not a salesperson.
Prefer to call? +44 7588 772736 · Not ready to talk? See it on your business first.
- 1Time
- 2Details
- 3Booked