AI Cost Calculators
Claude API Cost Calculator
Estimate Anthropic Claude API costs using input tokens, prompt caching, output tokens, Batch API, and monthly usage.
Enter Your Claude API Usage
Use average token values for one request, then enter the expected number of requests in one month.
Price used per 1 million tokens
Base input:$3.00
Cache write:$3.75
Cache read:$0.30
Output:$15.00
Estimated Claude API Cost
This estimate covers the token charges entered above.
Estimated monthly cost
Per request
$0.008550
Per day
$11.40
Per year
$4,104.00
Base input cost
36,000,000 tokens
$108.00
Cache write cost
0 tokens
$0.00
Cache read cost
80,000,000 tokens
$24.00
Output cost
14,000,000 tokens
$210.00
Requests: 40,000
Total input-related tokens: 116,000,000
Total output tokens: 14,000,000
* Important: Built-in rates checked June 19, 2026. Final charges may include other Anthropic services, taxes, discounts, retries, or usage not entered here.
Claude API costs can change with the model, input size, output length, prompt caching, Batch API use, and inference region. This calculator brings those values together so you can test a small launch, a normal month, and a higher-usage case before you build.
How the Claude API Cost Calculator Works
Choose a Claude model and enter the average token use for one request. Base input, cache writes, cache reads, and output tokens are calculated separately because each can use a different price.
Enter the number of requests you expect in one month. The calculator will show the cost per request, per day, per month, and per year.
You can also test Batch API pricing, US-only inference, or custom token rates.
What to Enter for a Useful Estimate
Use average values from a real request. Include the system prompt, user message, chat history, retrieved text, and tool instructions in your token estimate.
Only enter cache write and cache read tokens when prompt caching is part of your setup. Do not count the same tokens again as base input.
Test at least three cases: a small launch, a normal month, and a busy month. This gives you a clearer view of how the cost may grow.
Common Ways to Use This Calculator
- Estimate the monthly cost of a Claude chatbot.
- Compare Opus, Sonnet, and Haiku for one workload.
- Test the possible saving from prompt caching.
- Compare Standard API and Batch API estimates.
- Review the extra cost of US-only inference.
- Prepare an early Anthropic API budget.
Simple Claude API Cost Example
Imagine a support assistant with 40,000 requests per month. Each request uses 900 base input tokens, 2,000 cache-read tokens, and 350 output tokens.
Enter those values and choose Claude Sonnet. The calculator will show the separate input, cache-read, and output costs. You can then switch to Haiku or Opus without changing the workload.
Pricing and Estimate Notes
Built-in model prices were checked against Anthropic's official Claude API pricing documentation on June 19, 2026. Anthropic may change models, prices, or billing rules at any time.
The estimate covers token charges entered in the calculator. Web search, code execution, tool use, cloud platform premiums, taxes, and other services may add separate costs.
Always check the official Claude API pricing page before making a final budget or purchase decision.
Frequently Asked Questions
How is Claude API cost calculated?
The calculator multiplies base input, cache write, cache read, and output tokens by the selected Claude model rates. It then shows the estimated cost per request, day, month, and year.
What is the difference between a cache write and a cache read?
A cache write stores a prompt prefix for later use. A cache read uses that stored prompt again at a lower rate. The price depends on the selected cache duration and model.
Does Claude Batch API reduce the price?
Yes. Anthropic lists a 50% discount on input and output token prices for Batch API processing. The calculator can apply that reduction to the estimate.
What does US-only inference change?
For supported Claude models, US-only inference adds a 1.1 times price multiplier. Global routing uses the standard rate.
Are the results exact?
No. They are planning estimates. Your final bill may change because of price updates, tool use, web search, code execution, retries, discounts, taxes, or other paid services.
Can I enter my own Anthropic prices?
Yes. Turn on custom pricing and enter your own base input, cache write, cache read, and output rates per million tokens.