Cost planner

Describe your traffic once and compare the projected monthly bill across models, side by side. Then see at what volume a provisioned (PTU) reservation becomes cheaper than pay-as-you-go.

Workload

Chat assistant RAG Q&A Coding agent Batch summaries Embeddings

Requests per day Input tokens / request Output tokens / request Cached input

Sent as batch

Deployment type

Global Data Zone Regional

Compare

All models Retiring → replacements Popular Under $1 / 1M Partner models Clear

Projected monthly cost

InputCached inputOutput

Pay-as-you-go vs PTU

Monthly cost as daily volume grows, for the same prompt shape.

Model Peak ÷ average1× (flat)2×3×5×

Price sheet

List price per 1M tokens for the selected deployment type.

Download CSV

Media & audio estimator

Image, speech, video, document and search models are billed per image, minute, second, page or query rather than per chat request. Pick a group and enter your monthly volume.

Images Audio & speech Video Documents & search

Assumptions

Why some models aren't here

120 of 179 tracked models have pay-as-you-go prices (84 per token, 36 per image, minute, second, page or query). List prices are in USD from the Azure Retail Prices API for East US 2 (North Central US for classic speech and Whisper, which East US 2 does not sell), and from the public Azure Marketplace catalog for 15 partner models billed through Marketplace. Prices are checked daily, last updated 2026-11-01. They exclude tax, negotiated discounts and fine-tuning or hosting fees. A month is 730 hours. Per-image and per-minute figures for token-billed media models are estimates based on the editable assumptions. PTU sizing uses Microsoft's published tokens-per-minute figures and is an estimate: confirm it with the Foundry capacity calculator before you buy.