What your workload costs, per model

Enter the volume you actually run, then put any two models side by side. Prices that are already scheduled to change are carried through, so you can see what a switch is worth before and after 1 SEPT 2026. Nothing is sent anywhere — the maths runs in your browser.

Monthly volume
Input tokens620M tokens
Output tokens210M tokens

Cache-read pricing is excluded — most workloads don't hit it evenly, and including it would flatter models with aggressive caching. Input and output only.

Head to head
per month
vs
per month

Price only. statlyte records what models cost, not what they can do — a cheaper model is not automatically a substitute. Check capability before you switch.

Ranked for your volume — click a row to compare it

Alert me when any of these move