Anthropic Claude Code UK pricing is simpler than most developers assume: there is no separate Claude Code subscription, and the agentic coding tool comes bundled with the same Pro and Max plans you would buy to chat with Claude, or you can run it on pay-as-you-go API tokens. The catch is that Anthropic publishes every figure in US dollars and bills UK customers in dollars too, so working out what you will actually pay each month means converting carefully and adding VAT. This guide breaks down each route, converts the headline numbers into pounds at current rates, and explains who should pay for which tier, drawing on Anthropic’s official pricing page.
- Claude Code is included in the Claude Pro plan (USD 20/month, roughly £15) and both Max tiers; there is no standalone Claude Code subscription.
- Max costs USD 100/month (5x usage, around £75) or USD 200/month (20x usage, around £149), monthly billing only.
- On the API, Claude Opus 4.8 is USD 5 per million input tokens and USD 25 per million output tokens; Sonnet 4.6 is USD 3 and USD 15.
- Why it matters: Anthropic bills UK customers in dollars, so your real cost shifts with the exchange rate and UK VAT is added at checkout.
How Claude Code is sold, and why there is no separate price
The first thing to understand is that Claude Code is not a product you buy on its own. It is a terminal, IDE, desktop and browser coding agent that runs on top of an existing Claude account, and Anthropic’s own Claude Code documentation says plainly that most surfaces require either a Claude subscription or an Anthropic Console (API) account. That means there are two distinct ways to pay for it, and they are priced on completely different logic.
The first route is a flat-rate consumer or team subscription. Buy Claude Pro or Claude Max and Claude Code is bundled in at no extra charge, with your usage governed by session limits rather than a meter. The second route is the developer platform, where you generate an API key in the Console and Claude Code consumes tokens billed per million, with no monthly floor. A solo developer dipping in occasionally and a team running agents around the clock will reach for different routes, and the gap between them is the whole story of Claude Code value. If you are weighing this against rival assistants more broadly, our guide to choosing between Claude, Copilot and Gemini for UK work sets out the wider trade-offs before you commit to any one ecosystem.
One more structural point shapes the pounds-and-pence picture. Anthropic prices everything in US dollars and, for UK consumers, charges in dollars with VAT applied at checkout rather than publishing a separate sterling price list. So every figure below is an honest conversion of a dollar price, not a number Anthropic has set for Britain. We will flag that distinction throughout, because it is the single most common source of confusion when UK readers compare Claude Code against a rival that does quote in pounds.
The subscription route: Pro and Max in pounds
For most individual developers the subscription route is the obvious starting point. Claude Pro costs USD 20 a month billed monthly, or USD 17 a month if you pay USD 200 upfront for a year, according to Anthropic’s pricing page. At an exchange rate of roughly 0.745 pounds to the dollar in early June 2026, that monthly figure lands at about £15, with the annual deal working out near £149 for twelve months. Pro explicitly includes Claude Code, so a single £15-ish payment buys you both the chat product and the coding agent.
Max sits above Pro for heavier users and comes in two tiers. The entry Max plan is USD 100 a month and gives five times the usage of Pro; the upper tier is USD 200 a month for twenty times Pro usage. In sterling those are roughly £75 and £149 respectively. Unlike Pro, Max has no annual discount and is billed monthly only. The jump matters for Claude Code specifically because the bundled plans throttle on usage limits, so a developer who keeps hitting Pro’s ceiling mid-session is exactly the person Max’s higher multipliers are designed for. Our companion piece on Claude UK pricing across Free, Pro, Max and Team goes tier by tier if you want the chat-side detail.
There is also a Team route worth knowing about. Anthropic lists a standard Team seat at USD 20 per seat per month billed annually (USD 25 monthly), and a premium seat at USD 100 per seat annually (USD 125 monthly), which is roughly £15 and £75 per seat in sterling terms. Team adds shared administration and central billing, which matters once more than one or two developers are involved. For a UK small business deciding whether a paid plan is justified at all, our view on whether you actually need a paid AI subscription in 2026 is a useful sanity check before you start adding seats.
The API route: paying per token instead of per month
The alternative is to point Claude Code at an Anthropic API key and pay only for the tokens it consumes. Here the price depends entirely on which model you use and how much it reads and writes. Anthropic’s developer platform pricing sets Claude Opus 4.8, the most capable model, at USD 5 per million input tokens and USD 25 per million output tokens. In pounds that is roughly £3.70 and £18.60 per million respectively, before VAT.
Stepping down the range, Claude Sonnet 4.6 costs USD 3 per million input and USD 15 per million output, around £2.24 and £11.18, and is the model most production coding workloads settle on for its balance of speed and intelligence. Claude Haiku 4.5, the fastest and cheapest, is USD 1 and USD 5 per million, roughly £0.75 and £3.73. Token counts are hard to intuit, so a rough rule helps: one token is about four characters of English, and a meaty agent turn that reads a few files and writes a patch can run into tens of thousands of tokens. Two prompt-caching mechanics soften the bill considerably, which is where the API route earns its keep.
Prompt caching lets Claude reuse a previously processed chunk of context, such as a large codebase preamble, at a fraction of the input price. A cache read costs just 0.1 times the base input rate, so re-sending the same project context across many turns is far cheaper than it looks on the headline figure. There is also a Batch API offering a flat 50 per cent discount on both input and output for asynchronous, non-interactive jobs. For anyone scripting Claude Code in continuous integration, those two levers can halve or better the effective cost. The recent news that Claude Code reached a one billion dollar milestone as Anthropic bought Bun underlines how much real-world coding traffic now flows through exactly this metered route.
Subscription or API: which route should you choose
The decision turns on predictability versus precision. A flat subscription gives you a fixed monthly cost and no nasty surprises, which suits anyone who codes regularly and wants to stop thinking about the meter. If you would use Claude Code most working days, Pro at about £15 is almost certainly cheaper than paying per token for the same volume, and the moment you start hitting Pro’s session limits, the five-times headroom of Max for around £75 becomes the natural upgrade rather than switching billing models entirely.
The API route wins in two opposite situations. The first is very light use: an occasional automation, a weekend project, a script that runs Claude Code a handful of times a month. Paying a few pence in tokens beats committing to a £15 monthly subscription you barely touch. The second is heavy, automated, production use where you need fine control, model choice per task, prompt caching and batch discounts, and the ability to route cheap work to Haiku and hard reasoning to Opus. The middle ground, a steady solo developer, is where the flat subscription almost always works out simpler and cheaper.
One UK-specific wrinkle deserves attention. Because Anthropic bills in dollars, a subscription’s real sterling cost drifts with the exchange rate from one month to the next, and VAT is added on top at the standard rate for digital services bought by consumers. A £15 Pro plan is really a USD 20 plan that happened to convert to £15 this week. Budget against the dollar figure, not the pound estimate, if you want to avoid month-to-month surprises. Professionals in regulated fields should also read our notes for UK accountants using Claude and UK solicitors using Claude before pushing client data through any coding agent.
Claude Code versus GitHub Copilot for UK developers
The natural comparison for a UK developer weighing Claude Code is GitHub Copilot, and the two are priced on different philosophies. GitHub’s Copilot plans start with a free tier capped at 2,000 completions a month, then Copilot Pro at USD 10 a month (around £7.45) with USD 15 of monthly AI credits, Pro+ at USD 39 (around £29) with USD 70 of credits, and an individual Max tier at USD 100 (around £75) with USD 200 of credits. Notably, Copilot now exposes Claude models, including Opus and Sonnet, inside its agent mode and chat, so the underlying intelligence can overlap.
On a pure entry price, Copilot Pro at roughly £7.45 undercuts Claude Pro at roughly £15. But the comparison is not like for like. Claude Pro bundles the full Claude chat product and Claude Code’s deep multi-file agentic workflow under one payment, whereas Copilot’s credit-based higher tiers meter premium-model use against a monthly allowance. For a developer who lives in Claude Code’s terminal and desktop agents, the £15 buys a different, more autonomous experience; for someone who mainly wants inline completions in their editor, Copilot’s cheaper tier may be ample. Our Copilot UK pricing breakdown lays out the credit mechanics in full.
The honest answer is that price alone should not decide it. The two tools have different strengths in autonomy, multi-file reasoning and integration, and the model overlap means a Copilot subscriber can already reach Claude for the hardest tasks. We rate Claude Code higher for sustained, agent-driven work across a whole repository and Copilot stronger for fast in-editor assistance at a lower entry cost. For the broader assistant market, including ChatGPT’s developer tooling, our look at ChatGPT UK pricing in pounds rounds out the field.
Every price route at a glance
The table below pulls the headline numbers together so you can scan the routes side by side. The sterling figures are conversions of Anthropic’s published dollar prices at roughly 0.745 pounds per dollar in early June 2026 and exclude UK VAT, which is added at checkout. Treat the pound column as indicative; the dollar column is the contractual figure Anthropic charges.
| Route / model | Official USD price | Approx. GBP (ex VAT) | Claude Code included |
|---|---|---|---|
| Free plan | USD 0 | £0 | No |
| Pro (monthly) | USD 20/month | ~£15/month | Yes |
| Pro (annual) | USD 17/month (USD 200 upfront) | ~£149/year | Yes |
| Max 5x | USD 100/month | ~£75/month | Yes |
| Max 20x | USD 200/month | ~£149/month | Yes |
| Team standard seat | USD 20/seat (annual) | ~£15/seat | Yes |
| API: Opus 4.8 | USD 5 in / USD 25 out per MTok | ~£3.70 / £18.60 | Pay per token |
| API: Sonnet 4.6 | USD 3 in / USD 15 out per MTok | ~£2.24 / £11.18 | Pay per token |
| API: Haiku 4.5 | USD 1 in / USD 5 out per MTok | ~£0.75 / £3.73 | Pay per token |
A few patterns jump out. The free plan does not include Claude Code at all, so the realistic entry point is Pro at around £15. The Max 20x tier costs ten times Pro for twenty times the usage, which is unusually generous scaling for the heaviest users. And on the API, the spread between Haiku and Opus is fivefold on input and fivefold on output, which is precisely why routing cheap work to the small model and reserving Opus for hard reasoning is the standard cost-control tactic. For a sense of how Anthropic positions Opus itself, see our coverage of the Claude Opus 4.8 developer upgrade in the UK.
Where to sign up and manage billing in the UK
Because Claude Code is sold through Anthropic directly rather than UK retailers, there are no high-street sellers, delivery windows or returns to weigh up. Instead, the equivalent practical question is where you create the account and how billing is handled. Subscriptions are bought and managed at claude.com/pricing (last checked: 2026-06-07), where you can switch between Free, Pro, Max and Team and cancel or change tier from your account settings. Note that mobile app pricing can differ from web pricing because of app-store handling, so the web checkout is the reference point for the figures above.
For the API route, you sign up at the Anthropic Console (console.anthropic.com), generate an API key and point Claude Code at it; billing is postpaid against actual token usage with a card on file. The official install path is a one-line script per Anthropic’s docs, and the CLI prompts you to log in on first run. Whichever route you pick, the live, authoritative figures always sit on Anthropic’s own pricing and docs pages, so check them at the point of purchase since dollar prices and the exchange rate both move. If you are still deciding whether a coding-specific tool is the right spend, our roundup of the best AI writing assistant in the UK for 2026 covers the adjacent general-purpose options.
Anthropic Claude Code UK pricing: frequently asked questions
Is there a separate Claude Code subscription?
No. Claude Code is bundled into the Claude Pro and Max plans at no additional charge, and it can also be run on pay-as-you-go API tokens through an Anthropic Console account. There is no standalone Claude Code price, so the cost you pay is simply the cost of the plan or the tokens it consumes.
How much is Claude Code per month in pounds?
The cheapest plan that includes Claude Code is Claude Pro at USD 20 a month, which converts to roughly £15 at early June 2026 rates, before VAT. Max costs USD 100 or USD 200 a month, around £75 and £149. The free plan does not include Claude Code, so Pro is the realistic entry point for most individual UK developers.
Does Anthropic charge UK customers in pounds?
No. Anthropic publishes and charges in US dollars, with UK VAT added at checkout for consumers. There is no separate sterling price list, so every pound figure quoted by third parties, including this guide, is a conversion of a dollar price that moves with the exchange rate. Budget against the dollar amount to avoid month-to-month variation.
How much does the Claude Code API cost?
On the API you pay per token, by model. Claude Opus 4.8 is USD 5 per million input tokens and USD 25 per million output; Sonnet 4.6 is USD 3 and USD 15; Haiku 4.5 is USD 1 and USD 5. Prompt caching cuts cached reads to a tenth of the input rate, and the Batch API gives a flat 50 per cent discount on asynchronous jobs.
Is Claude Code or GitHub Copilot cheaper?
On entry price, Copilot Pro at USD 10 (around £7.45) undercuts Claude Pro at USD 20 (around £15). But they are not identical products: Claude Pro bundles the full chat product and an autonomous agentic coder, while Copilot’s higher tiers meter premium models against credit allowances. Copilot also now exposes Claude models, so the underlying intelligence can overlap.
Does the Max plan have an annual discount?
No. As of June 2026, Max is billed monthly only, at USD 100 for the 5x tier and USD 200 for the 20x tier. Only the Pro plan offers an annual option, at USD 17 a month paid as USD 200 upfront, and Team’s per-seat pricing is cheaper when billed annually. Always confirm current terms at checkout, as Anthropic changes plans periodically.
Which model should I use to keep costs down?
On the API, route routine edits and simple tasks to Claude Haiku 4.5, the cheapest model at USD 1 and USD 5 per million tokens, and reserve Claude Opus 4.8 for the hardest reasoning. Sonnet 4.6 is the sensible default for most production coding. Combining that routing with prompt caching is the standard way to control a Claude Code token bill.
Can a UK small business buy Claude Code for a team?
Yes. The Team plan adds shared administration and central billing, at a standard seat of USD 20 per seat per month billed annually (USD 25 monthly) and a premium seat at USD 100 annually (USD 125 monthly). Claude Code is included on Team seats just as it is on Pro and Max, so each developer gets the coding agent under one organisation account.
Our verdict
For the great majority of UK developers, we think the Claude Pro plan at around £15 a month is the right call: it bundles Claude Code with the full chat product, gives a predictable cost, and only needs upgrading to Max once you genuinely keep hitting session limits. We would steer light, occasional users to the API instead, where a few pence of tokens beats a monthly commitment, and we would point heavy, automated, production teams the same way for the model choice, prompt caching and batch discounts that flat subscriptions cannot match. Against GitHub Copilot, Claude Code is not the cheapest entry ticket, but for sustained agentic work across a whole repository we rate it the stronger tool, and the fact that Copilot now serves Claude models means the intelligence is not the deciding factor anyway. MTW score: 8/10 as a value and pricing-clarity assessment, not a hands-on lab test. The one thing that would change our recommendation is currency: because Anthropic bills in dollars, a sharp move in the pound could erode the value of a sterling-budgeted subscription overnight, so watch the exchange rate as closely as the feature list.
Related reading on MTW
Buyer action
Where to buy or check next
Use this as the final check before ordering a phone, changing network or trusting a headline monthly price.
Reader discussion
Leave a comment
Comments are moderated. Keep it useful, accurate, and on topic.