Skip to content
SmartCalcWorks

AI API Cost Calculator

Estimate what a large language model API will cost per request, per day and per month. Pick a model to load its published price, or type in your own prices. Then compare the same workload across every model in the table below.

Estimates only

Actual API costs may differ because of prompt caching, reasoning tokens, image or audio tokens, tool calls, long-context pricing, provider-specific billing rules and price changes.

Claude Sonnet 5.5: $2 input · $10 output per 1M tokens (USD)

Last verified: 18 October 2026 · Anthropic pricing(opens in a new tab)

US$
US$

Prompt + context + history.

Including billed reasoning tokens.

30 for every day, ~22 for weekdays.

Results update as you type.

Estimate (USD)

Estimated monthly cost

US$270.00

Input cost per request
US$0.004
Output cost per request
US$0.005
Total cost per request
US$0.009
Estimated daily cost
US$9.00
Tokens per month
60,000,000 in / 15,000,000 out
Based on Anthropic Claude Sonnet 5.5. Estimates only — check your provider’s billing for actual costs.

Compare models for this workload

Same token counts and volume, sorted by monthly cost. Add your own models — they’re saved only in this browser.

ModelProviderInput / 1MOutput / 1MMonthly
Claude Haiku 5.5 (prompts ≤ 100k tokens)Anthropic$0.1$0.5US$13.50
gpt-6-lunaOpenAI$0.1$0.5US$13.50
Gemini 3.5 Flash-LiteGoogle$0.3$2.5US$55.50
Gemini 3.8 FlashGoogle$0.75$3.75US$101.25
Claude Sonnet 5.5Anthropic$2$10US$270.00
gpt-6.1-solOpenAI$2$10US$270.00
Gemini 4 ArgonGoogle$2$10US$270.00
Gemini 3.1 Pro Preview (prompts ≤ 200k tokens)Google$2$12US$300.00
Claude Opus 5.5Anthropic$4$20US$540.00
Claude Fable 5.1Anthropic$10$50US$1,350.00
gpt-6-astraOpenAI$10$50US$1,350.00

Your models

Add a model with your own negotiated or newly published prices.

How the estimate is calculated

input_cost   = input_tokens  ÷ 1,000,000 × input_price_per_million
output_cost  = output_tokens ÷ 1,000,000 × output_price_per_million
per_request  = input_cost + output_cost
monthly_cost = per_request × requests_per_day × operating_days
Prices are per 1,000,000 tokens

Example

A support chatbot sends 2,000 input tokens and gets 500 output tokens back per request, 1,000 times a day, 30 days a month. At $2 input / $10 output per million tokens:

  • input: 2,000 ÷ 1,000,000 × $2 = $0.004
  • output: 500 ÷ 1,000,000 × $10 = $0.005
  • per request: $0.009 → per day: $9.00 → per month: $270.00

Getting realistic token numbers

  • Input tokens include everything you send: system prompt, conversation history, retrieved documents and tool definitions — not just the user’s latest message. In chat apps, history makes input grow with every turn.
  • Output tokens include the visible reply and, for reasoning models, any thinking tokens the provider bills as output.
  • Measure real traffic with your provider’s usage dashboard and plug those averages in here.

What this calculator leaves out

It only multiplies tokens by base text prices. It does not model cached-input discounts, batch discounts, long-context or regional multipliers, image, audio or video tokens, web-search or tool fees, fine-tuning, minimum spends or taxes. Several presets note where long-context pricing applies. All prices are in US dollars as published by the providers.

Planning to charge customers for an AI-powered service? The profit margin calculator helps you price above your API costs, and the Stripe fee calculator covers payment fees.

Official pricing pages

Frequently asked questions

What is a token?

A token is a chunk of text that a model reads or writes — often a short word or part of a longer word. As a rough rule of thumb, English text averages around three-quarters of a word per token, but this varies by model and language. Use your provider’s token counter or API usage reports for accurate numbers.

Why are output tokens more expensive than input tokens?

Generating text requires the model to produce tokens one at a time, which costs more compute than reading a prompt. Most providers price output several times higher than input, so long responses can dominate your bill.

Does this include prompt caching or batch discounts?

No. The calculator uses standard per-million-token prices. Prompt caching, batch processing and other discounts can reduce costs significantly; long-context surcharges, data-residency multipliers, reasoning tokens, tool calls and image or audio tokens can increase them.

Do I need an API key to use this calculator?

No. Everything is calculated in your browser from the numbers you enter. SmartCalcWorks never calls an AI API to produce these estimates.

How current are the model prices?

Each preset shows the date we last checked it against the provider’s official pricing page, with a link. Providers change prices often, so check the source — or enter your own prices — before making budget decisions.

Related calculators