Claude pricing in 2026 ranges from €0 to €180 a month for individual use: the Free plan costs nothing, Pro costs €15 a month billed annually or €18 billed monthly, Max costs €90 a month for five times the Pro capacity, and Ultra costs €180 for ten times the weekly usage of Pro. In the Netherlands, Anthropic shows these amounts in euros, excluding VAT. Programmatic use runs through the separate Claude API, which charges per token.
Which plan suits you depends on how often you hit a limit, not on which features you get on paper. If you rarely hit a limit on the free plan, stay where you are. If you get stuck in the middle of a task several times a week, Pro is the logical step. Max only pays for itself once you are regularly using up the five-hour window on Pro as well, and not before.
If you already work with Claude every day and still run out quickly, the answer more often lies in how you work than in a more expensive plan. If you learn to work with projects, set up longer tasks in one go and write more focused prompts, you use noticeably less capacity per result. The LearnLLM Claude Course sets out that way of working, so you only upgrade when you genuinely run short of capacity.
Claude pricing per plan
The plans differ mainly in usage capacity, not in the quality of the answers. With the same model, you get the same output on every plan; you pay for volume, extra tools and admin features. Which models you can use does vary by plan, and that matters more than the feature list.
| Plan | Price per month | Billing | Main inclusions |
|---|---|---|---|
| Free | €0 | n/a | Chat on web, mobile and desktop, web search, memory across conversations, file creation with code execution, connectors via remote MCP, Artifacts and Design, up to five projects; Sonnet and Haiku, no Opus |
| Pro (annual) | €15 | €180 a year in advance | Everything in Free, plus Claude Code, Cowork, Science, unlimited projects, access to Opus, Claude in Chrome and Claude for Microsoft 365 |
| Pro (monthly) | €18 | Monthly | Identical to Pro annual, without the 17 per cent annual discount |
| Max | €90 | Monthly | Five times the usage of Pro per five hours and per week, priority at peak times |
| Ultra | €180 | Monthly | Ten times the weekly usage of Pro, no five-hour limit for verified accounts (formerly Max 20x) |
| Team Standard | €18 per seat | Annual | More capacity than Pro per user, central billing, single sign-on, shared projects |
| Team Premium | €90 per seat | Annual | Five times the capacity of a Standard seat, otherwise identical features |
| Enterprise | $20 per seat plus usage | Annual only | SCIM, audit logs, compliance API, custom data retention, IP allowlisting |
All amounts in this table exclude VAT. Max and Ultra are billed monthly only. With monthly billing, Team seats cost €21.04 (Standard) and €105.23 (Premium) per seat. Anthropic shows Enterprise prices in dollars, even in the Netherlands.
Claude has no plan called Premium or Plus. The paid versions for individual use are Pro, Max and Ultra, with Team and Enterprise for organisations. If you search for Claude Premium or Claude Plus, you end up at Pro, Max or Ultra.
Since 22 September 2026, Opus 5.5 has been the current model in the Opus series, as the first model of the Claude 5.5 family. With that release, Anthropic raised the five-hour limits on Pro, Max, Team and seat-based Enterprise, without giving new figures. Since June 2026, the Fable line has sat above that series. Which Claude model you get by default on your plan directly affects how much you use, because heavier models get through your window faster. The context window goes up to one million tokens on all plans, depending on the model.
Why your invoice is higher than the price shown
The price Claude shows is a net price: Anthropic states for every plan that the amounts shown do not include tax. For a consumer in the Netherlands, 21 per cent VAT is added, which brings Pro to €18.15 a month billed annually and €21.78 billed monthly.
Businesses with a VAT number can enter it under Settings, Billing. In jurisdictions with a reverse charge mechanism for business digital services, Anthropic then stops adding VAT and you account for it yourself in your VAT return. For an office with five users, that immediately saves hundreds of euros a year in VAT you would otherwise pay up front.
Prices, currency and tax handling differ by region. The claude.ai/upgrade page always shows the amount that applies to your location, and it is the only place where you know for sure what you will pay.
Work out which Claude plan suits you
You see the plan that matches your usage and the amount per month and per year, with or without VAT.
Which payment methods does Claude accept?
You pay for Claude with a credit or debit card. Third-party methods such as PayPal and Venmo are not accepted, and ACH bank transfer is only available on Enterprise plans and for API accounts with monthly invoicing. The checkout page has no iDEAL button (iDEAL is the most common payment method in the Netherlands).
When a payment fails, the cause is rarely the amount. A billing address that does not exactly match the details your bank holds leads to a decline, as does an unfinished 3D Secure verification. For the Free plan, you do not need to enter a payment method at all.
Can you get a refund on a Claude plan within 14 days?
Within the European Economic Area, you can withdraw from a paid plan within fourteen days and request the refund directly in the app. Outside that period, payments are in principle non-refundable, except where the consumer terms or mandatory law say otherwise.
You can cancel at any time and your plan runs until the end of the paid period. Cancel at least 24 hours before the renewal date to avoid the next charge. Your conversations, projects and files stay in your account, even after it falls back to Free.
Can you use Claude for free?
You can use Claude for free on the Free plan, without entering a payment method. That plan gives you access to Claude on the web, iOS, Android and the desktop app, with web search, memory across conversations, extended thinking, connectors via remote MCP and the ability to create files with code execution.
Claude Code, Claude Cowork, Claude Science and the Opus models are not included in the Free plan, and projects are limited to five on Free. Since 16 September 2026, Anthropic has been merging Cowork and chat into a single Claude, starting with Pro and Max. Artifacts, with Claude Design, Slides and Docs, have been available on Free since 16 September 2026. If you want more than five projects or want to hand heavier reasoning tasks to Opus, you need at least Pro.
The main limitation is the usage limit, which works on a rolling five-hour window. So there is no daily reset at midnight: if you hit a limit at 14:00, capacity comes back around 19:00. For occasional use that works fine, but if you work with Anthropic’s Claude for several hours a day, you will keep running into that limit.
How much does Claude Pro cost?
Claude Pro costs €15 a month when you pay €180 annually in advance, or €18 a month when you pay monthly. The annual plan saves 17 per cent, which comes to about €36 a year. There is no separate Claude student discount on Pro: students pay the same rate, unless their university has bought access through Claude for Education.
For that amount you get at least five times the capacity of Free per five-hour window, plus Claude Code in your terminal, Cowork on web, desktop and mobile, Science, unlimited projects with their own context, access to the Opus models and Claude for Microsoft 365. Anthropic does not give a fixed number of messages or tokens, because the limit moves with the length of your conversations, the model you choose and the features you switch on.
On top of the five-hour window there is a weekly limit, and you share that capacity with Claude Code and Cowork. A long coding session in the terminal therefore drains the same pool as your conversations in the browser, which explains why some users get stuck sooner than expected, even on Pro.
The heaviest model, Fable 5.1, falls outside the standard capacity on Pro. If you want to use that model, you switch on usage credits and pay for that usage at the API rate on top of your plan. The tasks that genuinely need that heavy model are usually a smaller share of your work than it seems.
If you mainly work in a language other than English, the quality of Claude’s output in that language depends not on your plan but on the instructions you give it.
How much do Claude Max and Ultra cost?
Claude Max costs €90 a month and Claude Ultra €180 a month, both excluding VAT and without an annual discount. Max gives five times the usage of Pro, per five hours and per week. Ultra gives ten times the weekly usage of Pro and has no five-hour limit for verified accounts. Ultra is the new name for Max 20x, and existing subscribers were moved over automatically with no price change.
Besides that capacity, both plans give you early access to new features and priority at peak times. Since 7 October 2026, they include API credit: $100 a month with Max and $200 a month with Ultra, in dollars, as Anthropic lists them. The difference between Max and Ultra lies in volume, the five-hour limit and API credit, not in models or features.
On Max and Ultra, Fable 5.1 is available up to fifty per cent of your weekly limits, after which you can continue on usage credits at the API rate. For most individual users, Max is more than enough. Ultra is aimed at people who work for hours at a stretch every day, for example with large codebases, without being interrupted by a five-hour window.
Moving from Pro to Max is a capacity decision, not a feature decision. If you never hit a limit on Pro, Max buys you nothing you will use. If you come to a halt in the middle of a task several times a week, the jump is justified.
How much does Claude Team cost per seat?
Claude Team costs €18 per seat per month for a Standard seat and €90 per seat per month for a Premium seat, billed annually. Billed monthly, that is €21.04 and €105.23 per seat. Team is designed for organisations with two to 150 users.
The old minimum of five seats has gone. An office of two or three people can now take out a Team plan, which makes the step up from individual Pro accounts considerably smaller for small organisations.
The difference between Standard and Premium lies in capacity, not features. A Premium seat gives five times the usage of a Standard seat, both seat types get the same features, and you can combine the two within one organisation. Claude Code, Cowork, Design, Science and the Microsoft 365 integration are included with both.
On top of what Pro offers, Team adds central billing, single sign-on, enterprise search across your own organisation, admin controls for connectors and shared projects. Since 7 October 2026, a Team organisation also gets up to $500 a month in shared API credit, an amount Anthropic shows in dollars. By default, no model is trained on the content your organisation enters, and for firms with a duty of confidentiality, that outweighs every other argument.
How is Claude Enterprise priced?
Claude Enterprise is priced at $20 per seat per month plus actual usage, which is billed at API rates. Anthropic shows that seat price in dollars, even in the Netherlands. Billing is annual only, with a minimum of twenty seats. You can set up the plan self-serve or request a tailored proposal through sales.
This pricing model is very different from Team’s. With Team you pay a fixed amount per seat with a built-in usage cap; with Enterprise the seat price is low and the rest of the bill scales with the models and tasks your team runs. That is why admins can set spending limits per user and per organisation.
What you get for that price is all about governance. Enterprise adds role-based access control with fine-grained permissions, plus SCIM, audit logs, a compliance API, custom data retention policies, network access controls and IP allowlisting. For regulated sectors, there is also the HIPAA-ready version, which the lower plans lack.
From my background at Dutch financial institutions, I know that access control and demonstrable compliance are not nice-to-haves but hard requirements as soon as you work with sensitive data. If you mainly want predictable costs and collaboration within a fixed team, Team is enough. Only when audit logs, SCIM or your own retention periods are non-negotiable is Enterprise the route to take.
Does Claude Code cost extra?
No, Claude Code has no separate price, because Claude Code is included in all paid plans. There is no separate subscription; you pay nothing extra for it on top of Pro, Max, Ultra, Team or Enterprise.
Its usage does come out of the same pool as your chat usage. Tokens you use in the terminal count against the same five-hour window and the same weekly limit as Claude on the web. If you use Claude Code heavily for large codebases, you will reach the Pro limit sooner and drift towards Max.
There is an alternative for heavy sessions. Through a Console account you can switch to pay-as-you-go API credits, where you pay per token instead of running into your plan’s limit.
Claude API pricing per model
The Claude API charges per million tokens, split between input (your prompts and context) and output (the answer). The rates vary considerably by model.
| Model | Input per million tokens | Output per million tokens |
|---|---|---|
| Claude Fable 5.1 | $10.00 | $50.00 |
| Claude Opus 5.5 | $4.00 | $20.00 |
| Claude Sonnet 5.5 | $2.00 | $10.00 |
| Claude Haiku 5.5 | $0.10 | $0.50 |
Unlike the subscriptions, Anthropic publishes the API rates in US dollars. Opus 5.5 was released on 22 September 2026 at a rate twenty per cent lower than that of its predecessor Opus 5, which cost $5 input and $25 output. According to Anthropic, Opus 5.5 costs forty per cent less than Opus 5 on typical workloads, because the model also uses fewer tokens per task. Sonnet 5.5 was released on 28 September 2026 at the same rate as Sonnet 5, and according to Anthropic costs up to thirty per cent less per task for most work, because it works faster and needs fewer intermediate steps. Haiku 5.5 was released on 7 October 2026 at $0.10 input and $0.50 output for prompts up to 100,000 tokens; above that, $0.50 and $2.50 apply. Older models such as Opus 4.8, Sonnet 4.6 and Haiku 4.5 ($1 input, $5 output) remain available at their existing rates.
One detail affects your effective costs without the rate changing. Models from Claude 4.7 onwards, including Sonnet 5.5, use a newer tokenizer that produces about thirty per cent more tokens for the same text. If you migrate from Sonnet 4.6 or older, your bill per request may turn out higher while the rate per token stays the same.
If you need speed, Opus 5.5 offers a fast mode at double the rate, $8 input and $40 output, at up to 2.5 times the standard speed. If you need to keep processing within the United States, the data residency parameter adds a ten per cent surcharge on all token categories. Individual features have their own rates: web search costs $10 per thousand searches, and code execution is free when you use it together with web search or web fetch. Separately, you get 1,550 free hours of code execution per month, after which you pay $0.05 per hour per container.
Two mechanisms cut the bill substantially. Prompt caching brings the cost of repeated context down to one twentieth of the normal input rate for Opus 5.5 and Sonnet 5.5, and to one tenth for models including Haiku 5.5, which matters most for a system prompt that is identical in every request. Since 1 September 2026, Fable 5.1 and Mythos 5.1 go further still: reading from their cache costs only 0.025 times the input rate, or $0.25 per million tokens. For Opus 5.5, reading from the cache costs $0.20 per million tokens, sixty per cent less than for Opus 5. The Batch API gives a fifty per cent discount on all tokens for requests that can be processed within 24 hours.
In my Python integration with the Claude API, I combine both: Sonnet as the default for most pipeline work, Opus only for the more complex analysis steps, and prompt caching on the system prompt that stays the same in every run. Going from Sonnet to Opus doubles the rate, so in practice the choice of model matters more than the number of tokens. How to apply model choice, cost estimates and prompt caching yourself in an automation is covered in the LearnLLM AI Automation Course.
How much does Claude Fable 5.1 cost?
On the API, Claude Fable 5.1 costs $10 per million input tokens and $50 per million output tokens, two and a half times the rate of Opus 5.5. The model was released on 1 September 2026 as the successor to Fable 5 and is the heaviest publicly available model above the Opus family, intended for long-running agentic work, heavy reasoning tasks and analysis of large contexts. The context window is one million tokens and output runs to 128,000 tokens.
Alongside Fable 5.1 there is Claude Mythos 5.1, which runs on the same underlying model. Fable 5.1 has extra safeguards around cybersecurity and biology; Mythos 5.1 does not have those safeguards and is only available to a small group of vetted organisations: participants in Project Glasswing and vetted cyber defenders and life scientists through Anthropic’s Cyber Verification Program and Life Sciences Verification Program.
Availability of the Mythos class was interrupted at one point. On 12 June 2026, Anthropic suspended access to the predecessors Fable 5 and Mythos 5 to comply with export controls from the US Department of Commerce. On 1 July 2026, after the measures were withdrawn, it restored access to Fable 5 for users worldwide. On that date, Mythos 5 only came back for a group of US organisations.
On the subscription plans, Fable 5.1 is counted separately. Pro users and standard seats on Team pay for usage from the start through usage credits at the API rate. On Max and Ultra and on Premium seats in Team and seat-based Enterprise, the model is available up to fifty per cent of the weekly limits. On Enterprise with standard seats, the organisation has to switch on usage credits, and Enterprise with usage-based billing charges for Fable at API rates. With the 5.1 release, the base rates stayed the same and only reading from the cache was cut by 75 per cent, from $1.00 to $0.25 per million tokens. Anthropic measures about 25 per cent lower costs in normal use as a result, and up to 45 per cent in heavy agentic work, although independent measurements vary because the model produces more output tokens per task. If you reuse little context, you will notice little of that reduction, and for routine work Sonnet 5.5 and Opus 5.5 are often cheaper per completed task.
Claude pricing vs ChatGPT and Gemini
At entry level, Claude pricing is almost identical to the competition’s: Claude Pro and ChatGPT Plus are in the same price bracket as Google AI Pro (€21.99 a month). Google also has a cheaper plan, AI Plus at €4.99 a month, and AI Ultra at €99.99 or €219.99 a month. At that level, there is little to choose between them on price alone.
The difference lies in what you get in return. Claude positions itself on stricter instruction following and analysis of long documents, with a context window of up to one million tokens on all plans, depending on the model. ChatGPT offers a broader ecosystem of integrations, and Gemini integrates more deeply with Google Workspace.
When I switched from ChatGPT to Claude for most of my daily work, price played no part. What decided it was that Claude loses less context with long documents, sticks more strictly to instructions and makes up fewer details in factual questions.
If you are torn between the two, the best approach is to compare Claude and ChatGPT on the type of work you do every day.
At the Max level, both providers now have several heavier plans. Since 29 September 2026, OpenAI has offered three Pro tiers to rival Claude’s Max and Ultra, so very intensive users have a more expensive option on both sides.
When a Claude plan pays for itself
A Claude plan pays for itself the moment the free limit starts interrupting your work, not on the basis of time savings calculated in advance. So the question is not what you save, but how often you come to a halt.
Stay on Free as long as you ask something occasionally and rarely hit a limit. Move to Pro as soon as you have to wait for your window to reset in the middle of a task several times a week, because from that point you are paying with waiting time instead of money. Only consider Max when those interruptions come back every day on Pro as well.
For organisations, the sums are now straightforward. A Standard seat billed annually costs €216 a year, compared with €180 for an individual annual Pro plan, so about €36 extra per person per year. For that you get central billing, single sign-on, shared projects, enterprise search and admin features, plus the assurance that by default no model is trained on what you enter.
With Premium seats at €1,080 per seat per year, that calculation changes. That amount only makes sense for users who regularly hit the Standard limits, and in most offices there are only a few of them. A mix of both seat types is therefore almost always cheaper than one type for everyone.
If you want to learn the tool before taking out a plan, start on the Free plan. If you find you reach the capacity limit every day, that is the signal to upgrade.
Frequently asked questions about Claude pricing
Once you have chosen a plan, there are still practical questions about what Claude costs, mainly about buying extra usage and switching plans.
Can you buy extra usage when your Claude limit runs out?
Yes, on the paid individual plans you can buy extra usage when your Claude limit runs out, with usage credits that Anthropic bills separately from your plan at the standard API rates. You switch them on under Settings, Usage and buy credit in advance, optionally with automatic top-ups. A monthly spending cap keeps the extra costs under control. If you took out your plan through the mobile app, you buy usage credits through the web version.
Can you switch from Pro to Max mid-cycle?
Yes, you can switch from Pro to Max mid-cycle through Settings, Billing, and the upgrade takes effect immediately. You pay for a full Max period, minus a pro rata credit for the unused time on your Pro plan. Your billing period starts again on the date of the upgrade. If you paid for Pro annually and the remaining amount is higher than the Max price, you get the difference as credit towards later payments.
Can you share a Claude subscription with others?
No, Anthropic’s consumer terms do not allow you to share a Claude subscription with others. You may not share your login details with anyone or make your account available to someone else, and you remain responsible for everything that happens through your account. If you work with several people, a Team plan with a seat for each user is the route designed for that.
