LLM API cost calculator

How AI API pricing works

Every provider bills per million tokens with a lower price for input (your prompt, system instructions and attached context) and a higher price for output (the model's reply). Output is typically three to five times the input rate, so a chatbot that writes long answers costs more than one that reads long documents. Prices also differ by model tier: flagship models cost ten to thirty times more per token than small models, which is why routing simple requests to a cheaper model is the single biggest saving available.

Using this calculator

Enter your expected traffic, the average tokens in and out per request, and the input and output price per million tokens from the provider's pricing page. The defaults reflect a mid-tier model in 2026. The result shows the monthly bill, the daily run rate, the cost per request and the total tokens, with an approximate rupee conversion for Indian startups budgeting in INR.

Worked example

A customer-support bot handling 1,000 conversations a day with 800 input and 400 output tokens each, at $3 input and $15 output per million, costs about $8.40 a day or $252 a month, roughly ₹21,000. Moving to a model priced at $0.15 and $0.60 would cut that to about $11 a month, so the first optimisation to test is always whether a smaller model is good enough.

Ways to reduce the bill

• Shorten the system prompt; it is sent on every request.
• Use prompt caching for repeated context such as product catalogues or policies.
• Cap maximum output tokens for tasks that do not need long replies.
• Use batch APIs for non-urgent jobs; most providers discount them by about 50%.
• Route easy requests to a small model and escalate only when needed.

Frequently asked questions