
Claude by Anthropic has become one of the most popular AI assistants for writing, research and especially coding. But Anthropic's plans page has a lot going on: Free, Pro, Max, Ultra, Team, Enterprise, usage limits that reset every five hours, weekly limits, API rates per model and extra features like prompt caching. It is easy to get confused and pay for more than you need.
In this guide I break down every option in plain English. I checked Anthropic's official page while writing this, so the numbers are current for US buyers. You will learn what each plan costs, what you actually get, how Claude Code fits in, how API pricing works, and how to choose the cheapest option for the way you work.
Claude Pricing at a Glance

Here is the short version, taken from the official Claude plans page. Prices are in US dollars and do not include tax.
Free: $0. Limited usage every 5 hours and every week, chat on web, desktop and mobile, web search, file creation, memory and Artifacts.
Pro: $20/month billed monthly, or $17/month with the annual discount ($200 billed up front). Adds more usage, Claude Code, Projects, Research, more Claude models and Claude in Chrome.
Max: $100/month. 5x Pro usage every 5 hours and every week, priority access, Fable for up to half your weekly limit and $100 a month in API credits.
Ultra: $200/month. No 5-hour limit for verified accounts, 10x Pro usage every week and $200 a month in API credits.
Team: for 2 to 150 people. Standard seats cost $20 per seat per month billed annually ($25 monthly). Premium seats cost $100 per seat per month billed annually ($125 monthly).
Enterprise: $20 per seat per month billed annually, plus usage billed at API rates.
If you read older guides, you will see names like "Max 5x" and "Max 20x", where the 20x tier cost $200/month. Those guides are out of date, so do not plan around Max 20x numbers. The current page lists Max at $100 and Ultra at $200, with usage described as 5x and 10x of Pro. Always trust the official page over old blog posts, because Anthropic changes plans often.
Claude Free Plan: What You Get for $0
The free plan is a real product, not just a demo. You can chat with Claude on the web, desktop app and the iOS and Android apps, search the web, create files, run code, use memory across conversations, connect some apps and create Artifacts. Free users get the Sonnet and Haiku models and up to 5 Projects.
The catch is usage. Free has limited usage every 5 hours and every week, and you do not get Opus, Claude Code or Research. For everyday questions, emails and short writing tasks, Free is enough. For long documents, coding or daily work, you will hit the limit quickly.
Claude Pro: Is Claude Pro Worth $20 a Month?
Pro is the plan most people should start with. For $20/month, or $17/month when paid yearly, you get more usage every 5 hours and every week, access to Opus, Claude Code in your terminal, Projects to organize work, Research, Claude in Chrome, Claude for Microsoft 365 and Claude Science.
Anthropic says Pro gives you at least 5x more usage per 5-hour session than Free. You can also turn on usage credits to keep working at standard API rates when you reach a limit, and you can use Fable, Anthropic's top model, paid with usage credits.
Is it worth $20 a month? If you use Claude every day for writing, research or light coding, yes. It costs the same as ChatGPT Plus, and many writers and developers prefer Claude's style. If you only ask a few questions a week, stay on Free.
Claude Pro vs Max vs Ultra
Max and Ultra are for people who work with Claude all day. Here is how they compare:
Pro ($20): 1x usage. Good for daily use with breaks.
Max ($100): 5x Pro usage every 5 hours and every week, Fable included up to half your weekly limit, priority access at busy times, early access to new features and $100 a month in API credits.
Ultra ($200): everything in Max, no 5-hour limit for verified accounts, 10x Pro weekly usage and $200 a month in API credits.
Max and Ultra are billed monthly only. The monthly Claude Platform credits are a new and very useful extra: if you also build apps on the Claude Platform, those credits can cover a good part of your API bill.
My rule: start with Pro. If you hit the 5-hour limit several times a week and it slows your work, move to Max. Only go to Ultra if you run Claude Code or long agent tasks for many hours every day.
Claude Code Pricing: Which Claude Plan Do You Need?
Claude Code is Anthropic's coding agent that works in your terminal, IDE and the desktop app. It is included in all paid plans, and it shares the same usage limits as your chats. That means your coding sessions and your conversations draw from one pool.
Here is how I would choose for Claude Code:
Hobby projects and learning: Pro is enough. You will hit limits during long sessions, but you can wait for the limit to clear.
Daily professional coding: Max gives you room for long sessions and bigger codebases.
Heavy agent work all day: Ultra removes the 5-hour limit for verified accounts and gives you the biggest weekly allowance.
Teams: Team Standard or Premium seats with central billing and admin tools.
For very heavy coding, you can also switch Claude Code to pay-as-you-go usage through the Claude Console. Then you pay API rates per token instead of a flat subscription.
Hidden Claude Code Costs and Usage Limits to Watch
The biggest surprise for new users is the usage limits. Claude does not give a fixed number of messages. How fast you reach a limit depends on the model you use, how long your chats are and how large your files are.
What causes usage to spike? In my experience, these are the usual reasons:
Using Opus or Fable for everything: the top models use your limit faster than Sonnet.
Very long conversations: every new message includes the earlier context, so long chats cost more.
Large files and codebases: uploading big PDFs or letting Claude Code read many files uses a lot of tokens.
Agent loops: Claude Code running tests and fixing errors in a loop can burn through usage quickly.
What happens when you reach your limit? You can wait for it to reset, move to a higher plan, or on paid plans turn on usage credits to keep working at standard API rates. You can see where you stand any time in Settings, under Usage.
Claude Team and Enterprise Pricing
Claude Team is for groups of 2 to 150 people. It includes Claude Code and Claude Cowork for handing off longer tasks, Claude Design, Slides and Docs, enterprise search, connectors to company apps, central billing, single sign-on (SSO), admin controls and no model training on your content by default. You can mix and match seat types:
Standard seat: $20/seat per month paid yearly, or $25 monthly. More usage than Pro.
Premium seat: $100/seat per month paid yearly, or $125 monthly. 5x more usage than standard seats.
A Team Premium seat makes sense for heavy Claude Code and Cowork users, while Standard seats suit everyone else. Teams also get up to $500 a month in pooled API credits. Some features, like Claude Tag, which brings Claude into Slack, are aimed at team and business plans. Enterprise is for large businesses: $20/seat per month billed annually, plus usage billed at API rates. It adds SCIM, audit logs, custom data retention, role-based access, a compliance API, a HIPAA-ready offering and payment by ACH or invoice.
For enterprise buyers, the big questions are security and control. SSO, audit logs and admin controls for connectors matter more than the seat price for many companies. Enterprise also suits teams that run heavy workloads, because usage at API rates scales with real use instead of a fixed allowance per seat. If your company wants Claude across many workflows, ask sales for a tailored enterprise quote.
Claude API Pricing by Model

The Claude API is billed per million tokens (MTok). A token is roughly three quarters of an English word. You pay separately for input tokens (what you send) and output tokens (what Claude writes). Here are the current API rates for the latest models:
Fable 5.2: $8 input and $40 output per million. Built for long-running agents.
Opus 5.5: $4 input and $20 output per million. The daily driver for agentic coding and enterprise work.
Sonnet 5.5: $2 input and $10 output per million tokens. High performance for coding and agents.
Haiku 5.5: $0.10 input and $0.50 output per million tokens for prompts up to 100K tokens, and $0.50 input and $2.50 output for longer prompts. The fastest and cheapest model.
Older models are still available at their own rates. Haiku 4.5, for example, costs $1 input and $5 output per million, ten times more than Haiku 5.5 for short prompts. For example, Sonnet 5 costs $2 input and $10 output, Opus 5 costs $5 input and $25 output, and Fable 5 costs $10 input and $50 output per million. If you are on an older model, moving to the newest version can be both better and cheaper.
Getting Started With the Anthropic API: Console, API Key and Agent SDK
If you want to build with Claude, you start in the Claude Console. You create an account, add billing, create an API key and set spend limits. API usage is billed separately from any chat subscription, so a Pro plan does not include API calls unless you are on Max or Ultra with their monthly credits.
From there, you call Claude Opus, Claude Sonnet, Claude Haiku or Claude Fable through the Anthropic API. Developers who build agents often use the Agent SDK, and many connect tools through MCP, the Model Context Protocol, so Claude can work with files, databases and apps. Claude models are also offered through cloud platforms like Amazon Bedrock, Google Cloud Vertex AI and Microsoft Foundry, where pricing follows each cloud's own billing.
One thing to watch is the context window. Claude supports up to 1M tokens of context on some models, but sending huge context on every request gets expensive fast. Send only what the model needs.
Prompt Caching and Batch API: Two Ways to Spend Less
Two features can reduce your Claude API bill a lot:
Prompt caching: if you send the same long context again and again, the cache stores it. Cache writes cost a little more than normal input, but every cache hit is far cheaper. So like a big system prompt or a document, caching it makes repeated reads much cheaper. For Opus 5.5, a cache read costs $0.20 per MTok instead of $4 for normal input.
Batch processing: if your jobs do not need instant answers, the Batch API saves 50%. It is perfect for processing many documents overnight.
There are also some extras. Web search costs $10 per 1,000 searches, code execution includes 50 free hours a day per organization and then $0.05 per hour per container, and US-only inference costs 1.1x. Fast mode for Opus 5.5 and Fable 5.2 gives up to 2.5x faster speeds at 2x the standard price.
What Actually Drives Claude API Costs?

Three things control your real API bill: which model you use, how many tokens you send and receive, and whether you use caching and batch. Output costs five times more than input on most models, so long answers add up faster than long prompts.
Here is a simple example. Say your app sends 1 million input tokens and gets 200,000 output tokens a day on Sonnet 5.5. That is $2 for input plus $2 for output, so about $4 a day, or around $120 a month. The same job on Haiku 5.5 would cost a small fraction of that. If you cache the repeated part of the prompt and batch the work, the bill drops again.
My tips: use Haiku or Sonnet for simple tasks and Opus only where quality really matters, keep prompts focused, set spend limits in the Console and track usage per project.
Claude Code Subscription vs API: Which Is Cheaper?
For most individual developers, a subscription is cheaper than paying API rates. A heavy day of Claude Code on the API can easily cost more than a month of Pro. The subscription gives you predictable costs, and the Max and Ultra plans even include monthly platform credits.
The API makes more sense when you build your own apps, run automated jobs, need precise cost tracking per customer, or use the Batch API for large volumes. Many developers use both: a subscription for their own coding and the API for their products.
How Do Claude Costs Compare to ChatGPT and Gemini?
For consumer plans, Claude Pro, ChatGPT Plus and Google's AI Pro plan are all in the same range of about $20 a month in the US. The higher tiers differ more. Claude's Max and Ultra plans focus on more usage and API credits, while ChatGPT Pro and Google's top tier focus on their own premium models and features.
The best choice depends on what you do. Many writers and developers prefer Claude for long documents and code. ChatGPT is strong at images and voice, and Gemini fits well if you live in Google apps. Our guide on the best AI tools for writers compares them for writing work.
What Is the Cheapest Way to Use Claude?
Here are the cheapest ways, from free to paid:
Claude Free: $0, good for light use.
Pro with annual billing: $17 a month instead of $20.
API with Haiku and batch: very cheap for simple, high-volume tasks if you can code.
A shared plan: the lowest monthly cost if you can accept shared access.
At GroupBuyServices, you can use Claude through a shared plan together with ChatGPT and other AI tools for a small monthly fee. See bundles on our plans page.
Shared access has clear limits. It is a shared account, so do not paste private code, client data or personal information into it, and you do not get a subscription in your own name. It does not include API access for your own apps. For learning, writing and everyday questions, it is an affordable option. Our plans are non-refundable, so start with the 1 Day Trial if you want to test. You may also like our guide to premium AI tools for small business.
Claude Plans and Costs: Frequently Asked Questions
How much does Claude Pro cost?
Pro costs $20 per month billed monthly, or $17 per month with an annual subscription ($200 billed up front) in the US, before tax.
Is Claude Code included in Claude Pro?
Yes. Claude Code is included in all paid plans and shares the same usage limits as your chats.
What is the difference between Claude Max and Ultra?
Max costs $100 a month with 5x Pro usage. Ultra costs $200 a month with 10x Pro weekly usage, no 5-hour limit for verified accounts and more platform credits.
How much does the Claude API cost?
It depends on the model. Sonnet 5.5 costs $2 input and $10 output per MTok. Haiku 5.5 is the cheapest and Fable 5.2 the most expensive.
Do Claude usage limits reset?
Yes. Most accounts have a limit that resets on a rolling five-hour window, and paid plans also have weekly limits.
Final Thoughts
Claude's plans look complex, but the choice is simple for most people. Use Free for light use, Pro for daily work, Max if you hit limits often and Ultra only for all-day agent work. Developers building apps should use the API with prompt caching and batch processing to keep costs down. And if you want Claude on a tight budget for everyday tasks, our GroupBuyServices plans are a low-cost option to consider.