What this calculator answers
How much will an AI feature cost per month once real people use it? Enter the size of one request (in tokens, or paste a typical prompt), the length of the reply and how many requests you expect per day. The calculator returns the cost per request, per day and per month for every model with a published API price, cheapest first.
How to use it
- Size the input. Type the input tokens per request, or switch to Paste a prompt and paste one realistic request: the system prompt plus a real user message. Each model counts it in your browser: exactly where its tokenizer runs here, as a labelled estimate otherwise.
- Set the reply length. On the models listed here an output token costs between two and about eight times as much as an input token, so for chat-style replies this field often moves the total more than any other.
- Set the volume in requests per day. Monthly figures assume a 30-day month.
- Pick the model you plan to use. Its row is highlighted in the table, and if the provider sells a chat plan we have checked (ChatGPT Plus, Claude Pro or Max), the plan price is listed next to your API estimate as a reference for personal use.
What the estimate leaves out
The calculator uses each provider's standard, short-prompt rates. Prompt caching and batch processing can lower a real bill; very long prompts can raise it on models that charge more past a size threshold. Free tiers and taxes aren't included. Treat the result as a planning number, then compare it with your provider's usage dashboard once real traffic arrives.
Prompts are counted in your browser, so nothing you paste is uploaded. It's free, with no signup. For an exact count on Claude, Gemini, GPT-6 or Grok before you budget, use the Token Counter, which can ask the provider for its own number (opt-in: it sends the text to that provider). For ways to bring the total down, see how to cut LLM API costs.
FAQ
How is the monthly cost calculated? Input tokens per request times the model's input price, plus output tokens per request times its output price, gives the cost of one request. That is multiplied by your requests per day and by 30 days. Prices are each provider's standard rate per million tokens.
Where do the prices come from? From each provider's official pricing page, the same price list the Token Counter uses. The date they were last checked is shown under the table. Prices change, so check the provider before you commit to a budget.
What is not included? Cached-input and batch discounts, the higher rates some models charge for very long prompts, taxes and any free tier. Real bills can come out lower with caching or batching, or higher with long prompts.
Should I count reasoning tokens? Yes. Reasoning models bill their hidden thinking as output tokens, so set the output field to the reply plus the thinking you expect, not just the visible answer.
Is a ChatGPT or Claude subscription cheaper than the API? They pay for different things. ChatGPT Plus and Claude Pro cover a person chatting in the app and do not pay for API calls from your code. Claude Max and Ultra include a monthly API credit equal to their price, claimed by linking a Claude Console account; unused credit expires each month. The calculator lists these plans next to your estimate, for personal use.
Is my prompt uploaded? No. Pasted prompts are counted in your browser and nothing is sent or stored.