How much does it cost to run an AI feature?

At 25,000 requests a day with 1,200 input and 400 output tokens, a frontier tier runs about $7,200 a month, while a small model handling the same workload lands near $315.

What drives AI inference cost the most?

Output tokens. They usually cost four to five times as much as input tokens, so trimming response length cuts the bill faster than shortening prompts.

AI & Tech

AI inference cost estimator

Estimate what running a feature on a large language model costs per request, per day, per month and per year — and what the same workload would cost on a cheaper tier.

Tool

Turn request volume and token sizes into a real monthly model bill.

Requests per dayInput tokens per requestOutput tokens per request

Model tier

$3/1M in · $15/1M out

Monthly spend

$7,200

$87,600 a year at this volume

Cost per request$0.0096

Per day$240

Per week$1,680

Tokens per month1,200M

Same workload, other tiers

Frontier (monthly)$7,200

Mid-tier (monthly)$1,260

Small / fast (monthly)$315

How this is calculated

Cost per request is input tokens divided by a million times the input price, plus output tokens divided by a million times the output price. That is multiplied by daily request volume, then by 30 for a month and 365 for a year.

Prices are per million tokens and represent typical frontier, mid and small tiers rather than one named vendor.
Volume is flat across days; no caching, batching or prompt-compression discounts are applied.
Fine-tuning, storage, embeddings and egress are excluded.

Common questions

How much does it cost to run an AI feature?: At 25,000 requests a day with 1,200 input and 400 output tokens, a frontier tier runs about $7,200 a month, while a small model handling the same workload lands near $315.
What drives AI inference cost the most?: Output tokens. They usually cost four to five times as much as input tokens, so trimming response length cuts the bill faster than shortening prompts.