Claude Pricing 2026: What Does Claude Cost for Business?

Claude pricing is the question that lands on the owner’s desk right after the first colleague says “I already pay for Claude myself, can the company take it over?” Then the next one asks, and the one after that. Suddenly you are looking at Free, Pro, Max, Team, Enterprise and an API price list with five models and two prices each. This article puts every official Claude price in one place, as of October 2026, and shows what a team actually pays: per seat, per token, and without a seat for everyone.

What is Claude pricing?

Claude pricing is the set of plans and usage rates Anthropic charges for its AI model Claude: monthly seats for people who chat with Claude (Free, Pro, Max, Team and Enterprise) and per-token prices for software that calls Claude through the API. Seats pay for access and a usage allowance per person. The API pays for exactly what a program sends to the model and gets back. Enterprise combines both: a seat fee plus all usage at API rates¹⁵.

Claude for your whole team with Superkind

Many teams do not need a Claude seat for every person. They need the work done. Superkind is the AI employee for your team: you write to Superkind in Outlook or any email, Microsoft Teams, Slack or the Superkind web app like you would to a colleague, and it does the task in your programs. It works with AI models from Anthropic (the company behind Claude), OpenAI and Mistral, hosted and processed in Europe, GDPR compliant and with no model training on your data.

1. 1

   You write the task

   In Outlook, Teams, Slack or the web app, in plain words: “Compare the two forklift quotes.” No prompt course and no extra chat seat per colleague.

2. 2

   Superkind does it in your systems

   It takes the quotes from the inbox or SharePoint, uses your company knowledge (suppliers, prices, past orders) and follows your rules, such as “Quotes over EUR 10,000 need sign-off from management”.

3. 3

   Superkind asks before anything goes out

   You get the result with Approve or Reject. Nothing is sent, booked or changed without your OK, and your colleague can approve in Teams or Slack, wherever they work.

4. 4

   Next time it runs as a routine

   What you explain once keeps running, for example “Every Monday - send the quote comparison”. Every task adds to your Company Brain, and that knowledge stays when people leave.

How Claude pricing works

Anthropic sells Claude in five building blocks. Which one fits depends on who uses Claude: single people, a team, a large organisation or your own software.

1. Personal plans (Free, Pro, Max) - one person, one account. Free costs nothing, Pro adds more usage and Claude Code, Max adds 5x or 20x the usage of Pro¹³.
2. Team plan - shared billing and admin for 2 to 150 seats. You pick Standard or Premium seats per person and can mix both¹⁴.
3. Enterprise plan - a seat fee for access plus every token at standard API rates. Seats include no token allowance, admins set spend limits¹⁵.
4. API - no seats. Your software pays per million tokens (MTok) in and out, with a different price for each model².
5. Usage limits - every chat plan has limits that reset on a rolling five-hour window, and paid plans add weekly limits. More money buys more usage, not unlimited usage¹.

Claude plans compared

This table shows every Claude plan with a published price. Prices are US list prices per month, excluding tax.

As of October 2026. Source: claude.com/pricing¹, Max and Team details from the Anthropic Help Center³⁴.

Team adds up to $500 per month in pooled API credits for the organisation¹. Anthropic also lists an Education plan, but without a public price.

Claude API prices by model

If you build Claude into your own software, you pay per million tokens. The descriptions in the second column are Anthropic’s own⁶.

As of October 2026. Source: Anthropic Docs - Pricing², prices in US dollars per million tokens.

Prices that change the API bill:

• Batch API - 50% off input and output for jobs that can wait for the answer².
• Prompt caching - reading a cached prompt costs 5% of the input price on Opus 5.5 and Sonnet 5.5, 2.5% on Fable 5.1 and 10% on most other models. Writing to the cache costs 1.25x (5 minutes) or 2x (1 hour) the input price².
• Web search - $10 per 1,000 searches, plus the tokens of the search results².
• US-only processing - setting inference to the US costs 1.1x on all token prices for Claude 4.6 and later models²⁷.
• Long context - Claude 4.6 and later models (except Haiku 5.5) process the full 1 million token context at standard prices².

Examples: what teams pay for Claude

Six realistic set-ups, from the whole team without seats to your own API tool. All Claude figures are calculated from the official prices above, before tax.

1. The whole team works with Claude models through Superkind

Daniel in accounting emails Superkind: “Book the Möller invoice.” Superkind books it in DATEV with your rules, asks Daniel for approval, and his colleague Nadine sees the result in Teams. Nobody in the team needs their own Claude seat, and Superkind works with models from Anthropic, OpenAI and Mistral, processed in Europe. When the work grows, you build custom solutions and apps on the Superkind platform on top of what Superkind already knows, yourself or with the Superkind team. Superkind has no public price list: you start for free or book a demo.

2. The owner alone with Claude Pro

One person writes offers and emails with Claude. Pro costs $20 per month billed monthly, which is $240 a year, or $200 a year with annual billing¹. Fine for one person, but it is a personal account, without the central billing and admin of the Team plan.

3. A small team on Team Standard

Five colleagues get Standard seats. With annual billing that is 5 x $20 = $100 per month ($1,200 a year), with monthly billing 5 x $25 = $125 per month¹. Every new colleague adds another seat.

4. A mixed team with one heavy user

Four Standard seats and one Premium seat for the colleague who works with Claude all day: 4 x $20 + 1 x $100 = $180 per month with annual billing¹⁴. The Premium seat costs as much as five Standard seats.

5. Enterprise with the minimum seat count

20 seats x $20 = $400 per month, $4,800 a year, before anyone has written a single prompt¹. All usage comes on top at API rates. The bill moves with the model and the task, so set spend limits per user from day one⁵.

6. Your own API tool for incoming emails

Say a self-built tool sorts 10,000 emails a month, with 1,500 input and 300 output tokens each. That is 15 million input and 3 million output tokens. The monthly token cost²:

• Haiku 5.5 - 15 x $0.10 + 3 x $0.50 = $3.00
• Sonnet 5.5 - 15 x $2 + 3 x $10 = $60
• Opus 5.5 - 15 x $4 + 3 x $20 = $120

The tokens are cheap. Building the tool, connecting it to your mailbox and ERP, and keeping it running are not part of this price.

What Claude costs in practice (and what else you need)

The list price is only the start. Check these points before you sign anything.

• Tax and currency - Anthropic shows US prices excluding tax. Pricing, currency and tax handling vary by region, and claude.ai/upgrade shows the price for your location¹⁴.
• Usage limits - limits reset every five hours, with weekly limits on paid plans. Team members can buy prepaid usage credits to keep working after a limit¹⁴.
• Variable Enterprise bill - Enterprise seats include no tokens. Every token is billed at API rates, self-serve usage is prepaid in US dollars⁵.
• Where data is processed - on the Claude API from Anthropic, inference runs “global” or “US only”, and data at rest is currently stored in the US⁷. If your data protection officer requires processing in Europe, clarify this before you start.
• Model training - on Free, Pro and Max, you decide in the privacy settings whether your chats may be used to improve Claude⁸. Team excludes model training on your content by default¹.
• Seats versus use - a seat pays for one person’s access and usage, whether they open Claude daily or twice a month. Count who uses it every week before you buy seats for everyone.
• Superkind - there is no public price list. Start for free at app.superkind.ai, or book a demo and see how it runs in your systems.

How to start

You can decide this in an afternoon. Three steps:

Conclusion

Claude has clear prices: $0 to $200 a month for one person, $20 to $125 per Team seat, $20 per Enterprise seat plus tokens, and per-token API rates from $0.10 to $50 per million. The real question is whether you pay for seats or for work done. If your team needs tasks done in Outlook, Teams, Slack, DATEV or SAP, send Superkind one real task this week and compare it with what a seat would give you.

Frequently Asked Questions

Henri Jung

Co-founder of Superkind, the AI employee teams write to in Outlook, Teams and Slack. Henri helps companies, from small teams to the Mittelstand, hand their routine work to AI - without new tools, consultants or an in-house AI team.