Technology
AI Inference Cost
Estimate monthly AI inference cost from your own token counts, call volume and prices for each million tokens.
Add to workspaceYour numbers
Result dey update as you type; no Calculate button dey. All calculation dey run inside your browser.
Token prices here na examples, no be live prices for any provider or model. Replace dem with the prices wey apply to your own account before you use the result.
Monthly cost wey we estimate
—
—
How the result break down
| Di item | Value wey e get |
|---|
Monthly tokens na calls multiply by tokens per call. Dem dey apply prices per million tokens, as you enter am.
Quick look at the result
Dem dey scale the bars to match the bigger cost component and show whether input or output tokens dominate the bill.
Monthly cost as number of calls dey change
| Case wey you wan compare | Value wey e get |
|---|
Each row dey reuse your tokens per call and prices, na just the call count dey change, so cost dey follow volume increase straight.
Formula
Monthly cost = calls × (input tokens × input price + output tokens × output price) ÷ 1,000,000; prices na USD for each million tokens. Cost for each call = monthly cost ÷ calls.
Example wey show the calculation
Input na 2,500 tokens, output na 800 tokens, calls na 100,000, input price na $5 and output price na $20 for each million tokens. Input cost = 100,000 × 2,500 × 5 ÷ 1,000,000 = $1,250. Output cost = 100,000 × 800 × 20 ÷ 1,000,000 = $1,600. Total = $2,850 every month, or $0.0285 for each call. These prices na example only — enter your own plan prices to get figure wey you fit use.
Where e fit work and limits
All prices na examples wey you fit edit for USD per million tokens, no be current provider quote. Zero calls mean zero monthly cost; hypothetical cost for each call still follow your token and price assumptions. Caching, reasoning tokens, discounts, other modalities, taxes and rate limits need separate entries or estimates.
Questions wey people dey ask often
How una calculate am?
Each call cost na input tokens times input price plus output tokens times output price, divide by 1,000,000; that per-call cost times monthly call count na the monthly total.
Na real provider prices be the default prices?
No. Na example inputs be dem. Provider prices dey change and differ by model and tier, so enter the prices from your own plan or invoice before you use the result.
Why output cost dey often pass input cost?
Output prices per token usually higher pass input prices, and long outputs dey accumulate. The breakdown table show the split so you fit see which side you suppose optimise.
Wetin dem no include?
Cached or discounted tokens, batch pricing, retries, infrastructure, and staff time. If those ones matter for you, add dem separately instead make you treat this total as full cost of ownership.
How I fit estimate tokens per call?
Measure sample of real calls: record input and output token counts, find the average, and check again after prompts change. Averages from small short test calls usually understate production calls.
E go predict my bill exactly?
No. Na estimate from your averages and prices. Compare am with actual invoice for one month, then adjust token averages or prices until the model track reality.
Other tools wey relate to dis
- AI Carbon Estimate — emissions for the same number of calls, based on the energy assumptions wey you enter
- Savings from AI Workflow — whether the time wey AI tool save fit cover the monthly fee
- Calculator for electricity cost — electricity cost from the kWh usage and tariff wey you enter
Results na estimates; dem no replace advice from qualified professional.