Price one model call
It prices uncached input tokens, cached input tokens, output tokens, and any extra tool cost you add.
Free AI cost calculator
Your AI bill, one request at a time.
Token prices are hard to picture. Enter one real workload and see what each successful customer request costs—including repeated model calls, caching, and retries.
Step 1
Describe one request
Choose the AI model used for each call.
Processing speed
Batch is slower and cheaper
A retry is an extra attempt after something fails. For example, 0.1 retries means about 10 extra attempts for every 100 successful requests.
Step 3
Compare the same workload
Every row uses the token counts, cache rate, retries, and volume you entered. Lower cost does not mean equal quality or speed.
Comparing model estimates…
| Rank | Model | Successful request | Monthly | vs selected |
|---|
Plain-English method
How the calculator thinks
It uses three small steps. Nothing is sent to a server.
Build one user request
It multiplies the call price by the number of model calls needed to finish one user request.
Include failed attempts
It adds average retries, then multiplies by your monthly successful request volume.
Private by design: calculations happen inside your browser.
No API key is needed. We collect only daily aggregate counts of page views,
calculator engagement, model selections, Standard/Batch choices, and share-link copies.
Analytics never collect calculator inputs, token counts, calculated costs, IP addresses,
cookies, fingerprints, or persistent visitor IDs. We respect Do Not Track and Global Privacy Control.
Share links include your settings in the part after #; anyone you share a link with can see them.