AI Token Cost Calculator
How to Calculate Token Cost
Use billable token counts from the same workload and rates that apply to the same model and billing arrangement.
Choose a display currency; no exchange conversion is performed.
Match the billing category
Cached input, batch processing, fine-tuning, and tool usage may have separate rates. Use the rate that actually applies to each entered count.
What This Token Cost Estimate Includes
The result is the arithmetic charge for one entered token workload. It intentionally differs from the LLM API Cost Calculator, which scales average tokens across a recurring request volume and adds fixed monthly costs.
Token Workload Cost Examples
Large mixed workload
Token counts and entered rates
Estimated workload charge
Two million input tokens cost 6 and half a million output tokens cost 6.
One input token
Token counts and entered rates
Estimated workload charge
A per-million rate is divided by one million for a single token.
Keep precision for small jobs
A single request can cost less than one cent, so the calculation retains six decimal places even when a currency display may visually round.
Frequently Asked Questions
Still have questions about this calculation?
Try the CalculatorPer-Million Token Cost Formula
Providers commonly quote input and output usage per one million tokens. The calculator normalizes each count to that unit before adding the charges.
Formula
Input-token cost
input cost = input tokens ÷ 1,000,000 × input rate
Output-token cost
output cost = output tokens ÷ 1,000,000 × output rate
Total workload cost
total cost = input cost + output cost
Scientific Background
OpenAI's official pricing documentation is one example of provider documentation that quotes token usage by input and output category. This calculator does not copy any price: every rate is supplied by the user.