LLM API Cost Calculator
Estimate what an LLM feature will cost per request, per day and per month, and compare models side by side. Enter your token usage and request volume, edit the per-model prices to match your provider, and read the comparison chart. Everything runs in your browser.
New to this? Read the LLM cost calculator guide →
The prices below are editable examples grouped by tier, not a live price feed. Provider rates change often — replace any row with your model's current $/1M-token rates before relying on the totals.
| Model | Input $/1M | Output $/1M | Per request | Per day | Per month |
|---|
Monthly cost comparison
Monthly cost assumes 30 days. All maths runs locally in your browser — nothing is uploaded.
How to use the LLM cost calculator
- Enter the tokens a typical request sends (input) and receives (output), and how many requests you expect per day.
- Update the price rows to your models' current rates — every input and output cell is editable. Not sure how many tokens your prompt is? Measure it with the token counter.
- Compare the per-request, per-day and per-month cost across models, and read the chart to see the spread at a glance.
Why output tokens dominate
Most APIs charge more — often 3× to 5× — for output tokens than input tokens, because generating text is the expensive part. A chatbot that reads a long prompt but replies briefly is cheaper than one that writes essays. Splitting the estimate into input and output makes that trade-off visible, and shows why trimming a verbose system prompt saves less than capping the response length.
Getting a realistic number
Real traffic is bursty and prompts vary, so treat the total as a planning figure, not a bill. For a tighter estimate, measure the token count of a few representative prompts, average them, and remember that chat history and retrieved context inflate the input on every turn. Confirm current rates on your provider's pricing page before committing to a budget.
Frequently asked questions
How is the LLM API cost calculated?
Cost per request is (input tokens ÷ 1,000,000 × input price) + (output tokens ÷ 1,000,000 × output price). That is multiplied by your requests per day, and by 30 for the monthly figure. Every price cell is editable so you can match your provider.
Why are input and output priced separately?
Generating text costs more than reading it, so most providers charge 3–5× more for output tokens than input tokens. Splitting them shows why capping response length often saves more than trimming the prompt.
Are the default prices accurate?
They are editable examples grouped by capability tier, not a live price feed. LLM prices change frequently, so replace the rows with the current per-million-token rates for your models before trusting the totals.
How do I know my token counts?
Measure a few representative prompts with the token counter and average them. Remember that the system prompt, chat history and any retrieved context count as input on every request.
Is anything uploaded?
No. The calculator is pure arithmetic in your browser — no prompt, token count or cost figure leaves the page.