Claude Code Pricing UK 2026: What It Actually Costs

Claude Code has no standalone price. You cannot buy it on its own: it is bundled into every paid Claude plan, starting at £15 per month on an annual Pro subscription, or you can run it on metered API tokens and pay only for what you use. That single fact resolves most of the confusion around Claude Code pricing, and it is the part almost every comparison page buries beneath a plan table.

This guide covers what Claude Code actually costs a UK business in pounds, why the price you see may differ from the one your colleague sees, and how to work out whether a subscription or pay-as-you-go API access is cheaper for your usage. Prices were verified directly against Anthropic’s own pricing and developer documentation in August 2026.

How much does Claude Code cost in the UK?

Claude Code costs from £15 per month in the UK on an annual Pro plan, or £18 billed monthly. Both prices include 20% VAT. Heavier use moves you to Max from £90 per month, and team accounts start at £18 per seat.

The table below shows the UK prices as they render for a UK visitor. Every figure includes VAT, which matters because most Claude Code pricing articles quote US dollar prices that exclude sales tax and therefore understate what a British business actually pays.

Team and Enterprise accounts price Claude Code per seat rather than per person, and add shared billing and seat management on top of the same tool.

Claude Code is bundled, not sold separately

There is no Claude Code subscription. Anthropic includes it in the Claude plans themselves, so the question is never “how much is Claude Code” but “which Claude plan covers the amount of work I want to do”.

This trips people up because Claude Code is a command-line developer tool while the plans it ships inside are marketed as general Claude subscriptions. If you already pay for Claude Pro, you already have Claude Code. If your team pays for Team seats, every seat includes it. The only decision left is capacity: the plans differ by how much usage they allow before you hit a limit, not by whether the tool is switched on.

The second route is different in kind. Rather than a seat, you can point Claude Code at the Claude API and pay per token consumed. There is no monthly fee and no usage ceiling, but equally no cap on what you might spend. For teams already comparing AI coding tools we rate for UK businesses, this two-route structure is the main thing that separates Claude Code from tools sold on a flat per-seat basis.

Subscription or API: which is cheaper for you?

A subscription is a fixed monthly cost with usage limits. The API is metered with no limits and no ceiling. Light and predictable use favours the subscription; occasional or spiky use favours the API.

Anthropic publishes a worked example that makes the metered side concrete. A one-hour coding session on Claude Opus 5 consuming 50,000 input tokens and 15,000 output tokens costs $0.705 in total: $0.25 for input, $0.375 for output and $0.08 for session runtime. Turn on prompt caching so that 40,000 of those input tokens are cache reads, and the same session costs $0.525.

Set that against the $20 monthly Pro price and the rough crossover is around 28 hour-long sessions on the most expensive model per month, or roughly a session and a half every working day. Below that, metered access is cheaper. Above it, the subscription wins, and it wins by more the heavier you get. Treat this as a rule of thumb rather than a quote: it compounds a published plan price with a published worked example, and either input can move.

Claude Code API token pricing

API access is billed per million tokens, in US dollars, in every country. Rates run from $1 per million input tokens on the cheapest model to $5 on the most capable, with output tokens priced five times higher than input.

Three mechanisms cut that bill substantially. The Batch API halves both input and output costs for work that does not need an immediate answer. Prompt caching charges a cache read at a tenth of the standard input price, which is why the worked example above drops by a quarter once caching is on. Against those, two extras add cost: web search runs at $10 per 1,000 searches, and managed agent sessions add $0.08 per session-hour while running.

Why your price differs from the one you saw

Anthropic’s pricing page renders in local currency. A UK visitor sees pounds including 20% VAT; a US visitor sees dollars excluding sales tax. The same plan therefore appears at two different numbers depending on who is looking.

This is the single most common source of error in Claude Code pricing articles, and it is worth being precise about. Pro shows as $20 per month on the US render and £18 per month on the UK render. Those are not a straight currency conversion of one another: the UK figure includes VAT that the US figure excludes. Quoting the dollar price and converting it yourself will understate your bill.

The API behaves differently again. Token pricing is set in US dollars and billed in US dollars regardless of where you are, so there is no UK-specific rate and no VAT-inclusive equivalent. If you go the metered route, your effective cost in pounds moves with the exchange rate every month, which is a genuine budgeting consideration that a flat subscription removes. The same currency question shapes what AI tools cost UK businesses more broadly.

Has Claude Code pricing changed?

Plan prices have held through 2026, and one scheduled increase was cancelled. Claude Sonnet 5 was launched at an introductory $2 and $10 per million input and output tokens, due to rise to $3 and $15 on 1 September 2026. That rise will not now happen.

This matters more than a single rate change normally would, because Sonnet is the model most production coding work runs on. Anthropic has confirmed in its developer documentation that the introductory pricing is now the standard price, so teams that built forecasts around a September increase can revise them down. It is also a useful signal for anyone weighing the metered route: the direction of travel on token pricing has been downwards, not upwards.

What has changed is the model line-up rather than the price list. Anthropic ships new models regularly and retires older ones, and the rate attached to a given tier can move when the model behind it does. If you are budgeting more than a quarter ahead, check the current rates rather than relying on a figure from a guide published earlier in the year.

How Claude Code compares to Cursor, Copilot and Codex

The meaningful difference is the billing model rather than the headline price. Claude Code offers both a capped subscription and uncapped metered billing. Most competing tools sell a flat per-seat subscription only, which makes budgeting simpler but removes the pay-for-what-you-use option.

We have deliberately not quoted competitor prices here. Cursor, GitHub Copilot and OpenAI Codex all change their plans frequently, and we only publish figures we have verified against the provider’s own pricing page and recorded with a source and a date. What we can say structurally is that a dual subscription-or-API model suits teams with uneven workloads, while a flat per-seat price suits teams that want a predictable line in the budget and do not want to think about tokens at all. If you are weighing the wider assistant market rather than developer tooling specifically, how Claude compares with ChatGPT and Gemini covers that ground, and our Microsoft Copilot review looks at the Microsoft option in detail.

Which Claude Code plan should you choose?

Choose Pro at £15 to £18 per month if you code with it most days. Choose Max from £90 if you hit Pro’s limits. Choose Team from £18 per seat if you need shared billing. Choose the API if your usage is occasional or spiky.

One caveat worth holding on to: the subscription and the API differ in risk as well as price. A subscription stops you at a usage limit, while the API keeps billing. If you are pointing Claude Code at automated pipelines rather than sitting at the keyboard yourself, the metered route needs a spend alert behind it. Teams thinking about that kind of automation more broadly may find our guide to automating workflows without writing code useful, and if you would rather not make this call alone, an AI automation consultant can size it with you.

How we researched this guide

We verified every price in this guide directly against Anthropic’s own pricing page and developer documentation, recorded each figure with its source and date, and separated UK pound prices from US dollar prices rather than converting between them.

UK plan prices were taken from the UK render of Anthropic’s pricing page and include 20% VAT. API token rates were taken from the Claude platform documentation and are quoted in US dollars, which is the currency the API bills in worldwide. Where a figure in this guide is the result of a calculation rather than a published price, we have said so in the text. We do not quote competitor prices we have not verified ourselves. For the wider picture on what these tools cost, see our guide to the best AI tools for business.

Last updated: August 2026. Prices verified against Anthropic’s published pricing and developer documentation on 12 August 2026. UK plan prices include 20% VAT; API pricing is billed in US dollars and will vary in pounds with the exchange rate.

Alex Morgan

Business Technology Analyst

Alex specialises in business technology and connected systems, covering vehicle tracking, fleet management, AI tools, and dash cams for UK companies. With a background in telematics engineering, he analyses how emerging technology can improve efficiency, safety, and cost control — helping businesses make informed decisions about the tools that drive their operations forward.

Reviewed by

James Hartley

Technology & Innovation Reviewer