Deployment types change the price
| Deployment type | Where requests run | Price |
|---|---|---|
| Global | Any Azure region worldwide | Usually the lowest |
| Data Zone (US or EU) | Within the United States or the European Union | Higher than Global |
| Regional | One chosen region | Highest, for strict data residency |
| Provisioned (PTU) | Reserved capacity | Per hour, not per token |
A cost estimate that uses one list price for every call is wrong as soon as you use Data Zone or Regional deployments, which teams often need for data residency.
How Alye prices Azure OpenAI calls
- On the Models page you add each Azure resource and deployment once, with its type and region.
- Every call to that deployment is then priced at the matching Global, Data Zone or Regional price, and labelled with where it ran.
- If no regional price is known yet, Alye uses the global price and marks the cost as an estimate.
- For provisioned deployments, set your own effective per-token price to spread the hourly cost across services and customers.
Track it in two lines
Use the official OpenAI client with your Azure endpoint as usual; Alye records each call automatically.
# Python pip install alye import alye alye.init(service="support-reply") # set ALYE_API_KEY in the environment from openai import AzureOpenAI client = AzureOpenAI(azure_endpoint="https://<resource>.openai.azure.com", api_version="2024-10-21")
// Node.js
npm install alye
import * as alye from "alye";
import { AzureOpenAI } from "openai";
alye.init({ service: "support-reply" });
const client = alye.wrap(new AzureOpenAI({ endpoint: "https://<resource>.openai.azure.com", apiVersion: "2024-10-21" }));Alerts go to Microsoft Teams as well as Slack and email. See the docs and how LLM cost tracking works.