Claude Pricing UK 2026: Plans, API Costs and Enterprise Fees

Last updated: 7 Oct 2026.

Claude has no UK price list: Anthropic bills in US dollars only. As at 12 Sep 2026, Claude Pro costs about £14.80 a month (about £12.60 a month on annual billing), Max plans start at about £74 a month, Team starts at about £18.50 per seat per month with a two-seat minimum, and Enterprise is about £14.80 per seat per month plus usage billed at API rates. Prices are converted from US dollars at £1 = $1.35 on 11 Sep 2026; vendors bill in dollars and UK VAT applies on top. Prices checked 12 Sep 2026.

Claude plan pricing, checked 12 Sep 2026

Source: claude.com/pricing, read 12 Sep 2026. Anthropic publishes these figures in US dollars only; the pounds below are our conversion at the rate stated above.

For most UK mid-market teams, Team Standard at about £18.50 a seat is the sensible starting point. Enterprise is worth the conversation when governance, identity and access requirements drive it, rather than headcount alone.

Anthropic no longer publishes a separate headline price for the 20x Max tier, and it does not publish an Enterprise seat minimum on the pricing page. Where a figure is not published we have left it out rather than carry forward an older number. The page states that prices shown do not include applicable tax. Prices checked 12 Sep 2026; Anthropic revises pricing, and we re-verify this page quarterly.

What the Claude API costs

Rates from Anthropic's pricing documentation on platform.claude.com, checked 22 Sep 2026 with Claude Sonnet 5.5 added on 30 Sep 2026, per million tokens and converted to pounds.

Four things change the bill materially, and all four are in the same documentation.

• Sonnet did not get more expensive. Sonnet 5's rate was announced as introductory pricing to 31 Aug 2026 and is now its standard price. Anthropic's pricing page states that the increase scheduled for 1 September 2026 "will not occur", and Claude Sonnet 5.5 launched on 28 Sep 2026 at the same rate.
• Cache reads are much cheaper on the newest flagships. A cache hit costs 2.5% of the base input price on Fable 5.1 and Mythos 5.1 (about 19p per million tokens) and 5% on Opus 5.5 (about 15p) and, since 7 Oct 2026, on Sonnet 5.5 (about 7p), against the standard 10% on every other model. For an agent that re-sends the same long context all day, that is the single biggest lever on cost. For the design choices that decide an API bill, see how to reduce Claude API costs. If you are moving an existing integration from Opus 5, here is what changes with Opus 5.5.
• The Batch API is half price on both input and output, for work that does not need an answer in real time.
• Pinning inference to the United States costs more. On Claude 4.6 and later models, setting the API to US-only inference applies a 1.1x multiplier across input, output and cache. Global routing is the default and the standard price.

One date to diarise: Anthropic lists Claude Haiku 4.5 for retirement not sooner than 15 Oct 2026. If you have built cost-sensitive, high-volume work on Haiku 4.5, plan the move now rather than at short notice. Its successor, Claude Haiku 5.5, launched on 7 Oct 2026 at a tenth of Haiku 4.5's per-token price for prompts up to 100,000 tokens, and Anthropic says it costs around 75% less to run on average. Test before you switch: it counts the same text as more tokens, and it rejects some request settings that Haiku 4.5 accepted. For how it compares with OpenAI's cheapest model, see Claude Haiku 5.5 or GPT-6 Luna.

A token is roughly three-quarters of a word in English, so a million tokens is very roughly 700,000 to 750,000 words, in the region of a long novel. What that means for your own bill depends entirely on how much context each request carries (document length, conversation history resent with every message, and how many tool calls an agent makes), so we do not quote a typical cost per conversation here; it varies too widely to be a useful figure.

The current line is Claude Fable 5.1 and the limited-availability Claude Mythos 5.1 at the top, Claude Opus 5.5 beneath them, Claude Sonnet 5.5 as the general-purpose model and Claude Haiku 5.5, released 7 Oct 2026, as the cheapest. Claude Haiku 4.5 is still available, and Claude Opus 5, Claude Sonnet 5, Claude Opus 4.8 and Claude Fable 5 are legacy but still available. None of them is cheaper per token than the current model that replaces it, and Opus 5.5 is 20% cheaper than either Opus, so there is rarely a commercial reason to stay on them.

Paying for Claude from the UK

Anthropic bills every plan and every API call in US dollars. There is no sterling price list. Every pounds figure on this page is our own conversion at £1 = $1.35 on 11 Sep 2026, published so that you can budget in your own currency, and it will drift as the rate moves.

In practice, a UK buyer's card or bank applies its own exchange rate at the point of payment, plus any FX fee the card issuer charges. For business purchases, UK VAT at 20% applies on top; VAT-registered businesses should confirm reverse-charge treatment with their accountant, since Anthropic's UK VAT position can affect how the charge is booked.

Why the licence is the small number

Here is the part that surprises most buyers. The per-seat licence, whether Team or Enterprise, is rarely the figure that decides whether a Claude rollout succeeds. Buying licences gives your staff access to a chat window. It does not, on its own, connect Claude to your systems, teach it your processes, set guardrails for safe use, or get sceptical colleagues to change how they work.

Plenty of companies buy seats, see a flurry of early enthusiasm, and then watch usage quietly fade because nobody built Claude into the actual flow of work. The licence was paid for. The value was not captured. The gap between those two outcomes is deployment, and deployment is where the meaningful cost sits. Claude Code seats follow the same rule, and our guide to measuring AI coding tools shows how to tell whether developer licences are paying off.

This is the same pattern we describe in our breakdown of how much AI agents cost in the UK: the software subscription is a line item, but the engineering and change work around it is the investment that determines return. If you only budget for licences, you have budgeted for the cheap part.

What deployment actually involves

When we talk about deployment, we mean the work that turns Claude from a clever assistant into something your business relies on. In practice that covers several things.

• Connecting Claude to your data and tools. An assistant that cannot see your CRM, your documents or your ticketing system can only answer in general terms. Real value comes from grounding Claude in your own information, safely.
• Designing the use cases. Deciding which tasks Claude should handle, where a human stays in the loop, and what "good" looks like for each one.
• Setting guardrails. Defining what Claude is allowed to do, what data it may touch, and how outputs are checked. This is the governance layer that protects you.
• Building any custom agents. Where a workflow needs Claude to take action across systems, not just answer questions, that requires engineering.
• Enabling people. Training, documentation and support so that adoption sticks rather than fizzles after the first fortnight.

For a structured walk-through of the rollout itself, our Claude for Work deployment guide covers the practical sequence UK companies tend to follow.

Where SpotDev fits, and what it costs

SpotDev is an AI and digital transformation consultancy that builds real software, and is Certified in the Claude Partner Network, with deep specialism in Anthropic's Claude. We are not an Anthropic reseller and we do not sell Claude licences. Our credibility is hands-on engineering depth, with an in-house team that has delivered more than 300 technology projects and subcontracts nothing. We help UK organisations, typically from around £3m in revenue upwards, get Claude into productive use.

Because the deployment question is the one that actually matters, we price it as fixed packages rather than open-ended day rates. You know the cost before you commit, and the scope does not creep.

For most companies new to Claude, the £8,000 AI Foundations package is the honest answer to "what do I actually need". It is a contained, fixed-price engagement that gets a genuine rollout live, typically within two to three weeks, rather than leaving you with seats and good intentions. Our packages run from £3,500 to £45,000 depending on ambition, and the £5,000 Readiness Assessment credits 20% of its fee against whichever package follows.

For a single team of up to 15, AI Starter is the smaller starting point: set up and trained remotely in four weeks for £3,500. For a whole department or company, our Claude rollout service sets out how we set up Claude Enterprise or Claude Team, train your people in small online groups and measure the result at day 90.

The way to read the total budget is simple. The Claude licence is your ongoing access cost, paid to Anthropic per seat, in US dollars. The SpotDev package is the one-off engineering and enablement cost that makes those seats worth paying for. Keep the two figures separate in your planning and you will avoid the most common mistake, which is funding the licence and starving the deployment.

Prices checked 12 Sep 2026 against claude.com/pricing and platform.claude.com. Anthropic revises pricing; we re-verify this page quarterly.

Frequently asked questions

How much does Claude cost in the UK?

There is no UK price list; Anthropic bills everything in US dollars. Converted at the rate stated above, Claude Pro is about £14.80 a month, Team starts at about £18.50 per seat per month, and Enterprise is about £14.80 per seat per month plus usage billed at API rates. Your bank or card applies its own exchange rate and any FX fee, and UK VAT at 20% applies to business purchases.

What is the minimum number of seats for Claude Team?

Two seats, and the plan runs up to 150 users. Team Standard is about £18.50 per seat per month (about £14.80 on annual billing); Team Premium is about £92.50 per seat per month (about £74 on annual billing).

How much is Claude Enterprise?

The seat licence is about £14.80 per seat per month, plus usage billed at standard API rates, billed annually. Anthropic does not publish a seat minimum for Enterprise on its pricing page, so treat any figure you see quoted elsewhere as unconfirmed. Because usage is billed separately, the total monthly cost depends on how heavily your team uses Claude, not on the seat fee alone.

Is Claude cheaper on annual billing?

Yes. Pro drops from about £14.80 to about £12.60 a month, Team Standard from about £18.50 to about £14.80 per seat per month, and Team Premium from about £92.50 to about £74 per seat per month. Enterprise is billed annually, so there is no monthly comparison to make.

What does it cost to deploy Claude properly?

The licence, whichever tier you choose, is usually the smaller part of the total cost. The larger cost is deployment: connecting Claude to your systems, designing use cases, setting guardrails and training your people. SpotDev prices this as fixed packages from £3,500 to £45,000, with a £5,000 Readiness Assessment (20% credited against a later package) for companies that want a costed plan first. See our AI implementation packages for the full breakdown.

Request a Quote to discuss a Claude rollout for your business.