Start with the output ratio#
A million input tokens and a million output tokens are different products. With Qwen3 32B, 800,000 input and 200,000 output tokens cost $0.16 at the listed rates. The same total with an 80% output mix costs $0.19.
Count the whole workflow#
Include retrieved documents, repeated system prompts, tool results and retries. Agent loops can generate many requests for one visible answer.
Use your trace#
Take a representative sample, total actual usage fields and scale by expected volume. The calculator uses the same rates as the catalogue.