Free tool

LLM Cost Calculator

The per-token price is not the cost of your feature. Enter your real usage once, then compare models side by side to see the monthly bill for each. Everything runs in your browser. Nothing is sent anywhere.

Your usage

Monthly active users

Sessions / user / month

Input tokens / session

Output tokens / session

Retry rate (%)

Cache-hit rate (%)

Agent multiplier (1 = one call, up to 10 = loop)

Compare models

Estimated monthly cost, cheapest first

Model	Input / 1M	Output / 1M	Monthly cost	vs cheapest

Cached input is billed at about 10% of the normal rate, so a higher cache-hit rate cuts the input line. Output has no cache discount. This is an estimate, not a quote.

Prices verified 7 July 2026 against each provider's official pricing, and updated periodically. They move often, so confirm against the provider before you commit a number. Claude Sonnet 5 shows its standard $3 / $15; it is running an introductory $2 / $10 through the end of August 2026.

Surprised by the spread?

Picking the right model for each call, and pulling the levers that cut the bill (caching, batching, trimming context), is the difference between the top of this table and the bottom. I size AI features against real numbers as part of AI Enablement.

Book a free 30-minute call

Want the reasoning behind the formula? Read what an LLM feature really costs.