What Claude AI Costs a Business: Plans, Models and Pricing

Short answer: Claude is an AI assistant made by Anthropic. A business can use it in two ways: as an app with a monthly plan (from $0, paid plans from $20 a month), or through the API inside its own chatbot or workflow, paid per use. Anthropic's newest model, Claude Fable 5.2, was released on 9 November 2026.

All prices below are list prices in US dollars, taken from Anthropic's own pages on 11 November 2026.

Key points

• Claude is sold in two ways: an app with monthly plans, and an API that developers build into other software.
• App plans run from free to $200 a month for one person, and $25 per seat a month for a team.
• There are four current models. The cheapest, Claude Haiku 5.5, costs $0.10 per million input tokens. The newest, Claude Fable 5.2, costs $8.
• Anthropic itself recommends starting with Claude Opus 5.5 for most work, not the top model.
• For a business chatbot the model fee is usually the smallest cost. The build and upkeep cost more.

What Claude AI is and what it is used for

Claude is a family of AI models and the app built on them. You type or upload something, and it writes, summarises, analyses, answers questions or writes code. It competes with ChatGPT from OpenAI and Gemini from Google.

For a business owner there are two different products behind the one name.

• The Claude app. A person on your team opens Claude on the web, desktop or phone and works with it. You pay per seat per month.
• The Claude API. A developer connects a Claude model to your website chat, WhatsApp number, CRM or internal tool. Your customers or staff never open the Claude app. You pay for what the system uses.

Most confusion about cost comes from mixing these two.

Claude AI pricing plans: price per month

Source: Anthropic's pricing page. Team also has a premium seat at a higher price, and there is an Ultra plan at $200 a month for very long tasks.

One point matters for client data. Anthropic's plan table lists model training as "None by default" on Team and Enterprise, and as "Opt-out" on Free, Pro and Max. If your staff paste customer details into Claude, use a Team plan or switch the training setting off first.

Claude AI models compared: API pricing per token

Through the API you pay per million tokens. A token is a small piece of text, roughly a part of a word. Input is what you send, output is what the model writes back.

Three details from Anthropic's price page that change the bill:

• Haiku 5.5 is cheap only for prompts up to 100,000 tokens. Above that, the same request costs $0.50 and $2.50.
• Batch processing, where you do not need the answer at once, is 50 percent cheaper on input and output.
• Prompt caching lowers the price of text you send again and again, such as your product catalogue or policy.

Anthropic's own advice on choosing: "For most coding and knowledge tasks, we recommend that users start with Claude Opus 5.5". The top model is not the default choice, even for its maker.

There is also Claude Mythos 5.2, released the same day as Fable 5.2. It is available only to vetted organisations, so it is not an option for a normal business.

A worked example: 1,000 customer chats

Say your support bot handles 1,000 chats a month. Each chat sends about 2,000 tokens to the model (the customer's messages plus your instructions and product details) and gets about 500 tokens back. That is 2 million input tokens and 0.5 million output tokens.

This is our own arithmetic on list prices with assumed chat sizes. Your real numbers depend on how long your chats and instructions are.

The lesson is not that AI is nearly free. The model fee is usually the smallest line. The real cost of a business chatbot or agent is the build, the connections to your systems, testing, and upkeep. We break that down in AI Automation Cost 2027.

Claude AI for small business: app or custom build

Use the Claude app when the work is done by a person: drafting emails and proposals, summarising long documents, analysing a spreadsheet, preparing for a meeting. A Pro or Team seat is enough. No developer is needed.

Use the API in a custom build when the work should happen without a person starting it: answering customers on your website or WhatsApp, reading incoming invoices, routing leads into your CRM, running a multi step task from start to finish. That is where AI chatbots and AI agents come in.

A common middle path is a workflow tool such as n8n or Zapier with a Claude model plugged in. See n8n vs Zapier vs custom automation.

If AI agents are new to you, start with getting started with AI agents.

Claude AI vs ChatGPT and Gemini: how a business should choose

People search for a winner. For a business there is rarely one. Claude, ChatGPT and Gemini all come as an app and as an API, and their small and mid level models are priced in a similar range. A fair way to choose:

1. Collect 20 real tasks from your own work: real customer questions, real documents, real emails.
2. Run the same tasks through each tool and let the person who does that job today judge the answers.
3. Check the data terms of the exact plan you would buy, not the brand in general.
4. Check what it connects to. The tool that works with your CRM, inbox or store may matter more than a small gap in answer quality.

A well built system can also switch models later, so this is not a decision for life.

What changed recently

• 9 November 2026: Claude Fable 5.2 released at $8 and $40, down from $10 and $50 for Fable 5.1.
• 21 October 2026: Claude Duet, an app where Claude works as a proactive agent, opened as a research preview for Pro and Max plans.
• 7 October 2026: Claude Haiku 5.5 released as the cheapest model.
• 22 and 28 September 2026: Claude Opus 5.5 and Claude Sonnet 5.5 released.

The pattern for a buyer: the newest top model costs less than the one before it, and there is now a very cheap model built for high volume jobs.

Limits to keep in mind

• Claude can state wrong things with confidence. Anything that goes to a customer or into your accounts needs a check or a clear rule for handing over to a person.
• It only knows your business if you give it your information, through instructions, documents or a connection to your systems.
• If your bot talks to people in the EU, the EU AI Act requires that they are told they are talking to an AI, unless that is obvious.
• Prices and models change every few weeks. Do not build a budget on a number from an old article, including this one. Check the price page.

Next step

If one person on your team needs help with writing and analysis, start with a Pro or Team seat this week. No project is needed.

If you want Claude or another model working inside your website chat, WhatsApp or back office, the first step is to pick one process and measure it. That is where our Automation Audit starts: $700, 7 working days, credited to the build if you go ahead. See pricing or tell us what is stuck.

Sources and notes

• Anthropic, Claude plans and pricing: https://claude.com/pricing
• Anthropic, API model pricing: https://platform.claude.com/docs/en/about-claude/pricing
• Anthropic, Claude Fable 5.2 and Claude Mythos 5.2 announcement, 9 November 2026: https://www.anthropic.com/claude-fable-and-mythos-5-2
• Anthropic, Claude release notes: https://support.claude.com/en/articles/12138966-release-notes

Prices checked on 11 November 2026. Disclosure: Stuck Media Solutions uses Claude in its own work. We are not affiliated with Anthropic, and every Claude price in this guide comes from Anthropic's published pages.