Claude API cost calculator

How much does using Claude through the API really cost? Pricing is not based solely on message count, but primarily on input and output token counts, the selected model and additional options enabled in the request.

Choose a model, enter your estimated token count and usage volume, and the calculator will show the estimated cost in dollars.

Anthropic also uses the term usage credits for a monetary balance charged for usage. Credits are not a token unit, so there is no fixed ratio equating one credit to a particular number of tokens or requests.

Claude API cost calculator

Prices last checked on September 23, 2026

Advanced options
Advanced options

$10 per 1,000 successful searches ($0.01 per successful search)

Designed for asynchronous processing of requests that do not require an immediate response.

Estimated cost per request
Estimated total cost
Input token cost
Output token cost
Estimated cost for 100 identical requests
Estimated cost for 1,000 identical requests

This calculation is an estimate only, based on list prices and the data entered. Actual charges are determined by usage recorded by Anthropic and may vary depending on tokenization, caching, tools, thinking settings, API settings and pricing changes.

How is Claude API pricing calculated?

Claude API usage is priced primarily by the number of tokens processed in each request. Input tokens include instructions, messages, documents and other information sent to the model. Output tokens are the content the model generates in response.

Each model may have different input and output prices. Options such as Prompt Caching, Message Batches, Web Search and special rates for certain models can also change the cost.

The calculator breaks costs into components so you can see both the total and the contribution of input, output and additional options.

What are Claude usage credits?

Usage credits are not a fixed unit of tokens. They are a monetary balance that can be charged for usage according to the rates applicable to the actions performed.

You therefore cannot equate one credit to a fixed number of tokens or requests. The same balance can cover different amounts of usage depending on the model, input and output length, and additional features enabled.

What is the difference between input and output tokens?

Input tokens are the tokens sent to the model. They can include the prompt, conversation history, system instructions, documents and other information included in the request. Output tokens are the tokens the model generates in its answer.

In most API models, output token pricing differs from input token pricing. Two requests with the same total token count can therefore cost different amounts if their input-to-output ratios differ.

Why can actual costs differ?

The calculator provides an estimate based on the numbers entered and list prices checked on the date shown on the page. In practice, the exact token count is determined after the content is tokenized. Conversation history, system instructions, tools, extended thinking, documents and additional features can affect token counts and charges.

Anthropic may also change prices or terms after the verification date. For actual billing, the data in your Anthropic account is authoritative.

What is Prompt Caching?

Prompt Caching lets you cache recurring parts of the input so the same content does not need to be processed again in the same way for each eligible request. Pricing may distinguish between cache writes, cache reads and standard input.

This can be useful when the same large context recurs across many requests, such as long system instructions, a fixed document or a repeatedly used dataset. Actual savings depend on usage patterns, the amount of cached content and how often the cache is reused.

What is Message Batches?

Message Batches API is designed for asynchronous processing of a set of requests that do not require an immediate response. According to the pricing checked for this calculator, eligible Message Batches usage receives a 50% discount on token-priced cost components.

This can suit document processing, classification, content generation or other tasks that can run in the background. Fixed tool charges such as Web Search do not receive the Batch discount unless Anthropic's pricing explicitly states otherwise.

Does a Claude Pro or Max subscription include API usage?

A Claude product subscription and billing for Claude API usage are separate usage arrangements. Do not assume that a Claude Pro or Max subscription provides an API allowance as part of its monthly price.

If you use the API, check billing settings and the balance in the relevant account, along with Anthropic's current plan terms. This page's calculator estimates API costs using the pricing configured in it.

What about Claude Code?

Claude Code is a developer tool, but billing can depend on the plan and how the user connects to the service. This calculator estimates costs using its configured Claude API rates. It is not designed to reproduce every type of charge that may arise when using Claude Code through another plan.

To check actual Claude Code billing, compare the estimate with your account information and Anthropic's current documentation.

Do images and PDFs cost more?

Images, PDF documents and other content types can affect the amount of input the model needs to process. Costs depend not just on the number of files, but on how the service processes them and the token counts or billing components actually generated.

Do not assume that every image or PDF has a fixed surcharge. If you plan extensive document or image usage, check Anthropic's current documentation and measure real requests.

Does Web Search cost extra?

Yes. Using Web Search through the API may incur an additional fixed charge beyond token costs. According to the pricing checked for this calculator, it costs 10 dollars per 1,000 successful searches, or 0.01 dollars per successful search.

The calculator adds this cost separately after calculating token components. If Anthropic changes Web Search pricing, update the calculator's pricing file rather than hard-coding the price in interface components.

When were prices checked?

The calculator's prices were last checked on September 23, 2026. API prices and terms of use can change.

Before a significant budget decision, compare the calculator's result with Anthropic's current official price list.

Official sources