DCA Model Gateway

One endpoint, and a key for each of your customers or applications, in front of the model deployments you already run in your own Azure AI Foundry. Billed through Microsoft Marketplace on your Azure invoice.

Subscribe in the Azure portal Marketplace listing Usage and support

From purchase to first routed request

Step 1

Open DCA Model Gateway in the Azure portal, pick the Team plan (or your private Partner Edition plan) and the Azure subscription to bill, and complete the purchase.

Step 2

Activate

Select Configure account now on the new SaaS resource, sign in with the Microsoft Entra work account you bought with (single sign-on, no new password), accept the privacy notice and select Activate. Your first gateway key is shown once: store it in your secret manager.

Step 3

Connect your Foundry

Add the DCA Model Gateway Connector to your Microsoft Entra tenant and give it a role on your Foundry resource; our custom role allows model calls only. Enter your endpoint, map your deployments to routes and select Save and test. No Foundry keys are shared or stored.

Step 4

Call the gateway

Point your OpenAI or Anthropic SDK at the gateway with your key. x-dca-route (cheap, auto, reasoning, code, frontier, oss-only) picks one of your mapped deployments. Every response reports the deployment used, the data boundary applied and your remaining token budget. Add a key for each customer under Tenant keys.

What is metered, and what is not

Model inference is billed by Microsoft on your own Azure subscription, as with a direct call to your deployment; the gateway does not resell it. DCA Model Gateway charges a monthly plan fee through Microsoft Marketplace. Routed tokens beyond the included amount are metered per 1M, separately for open-weight and frontier models, and so is each active tenant beyond the included number. Requests the gateway refuses are not counted.

Team

$199 /month

Up to 25 keys. Included each month: 5 active tenants, 200M open-weight and 25M frontier routed tokens.

Partner Edition

$499 /month

Up to 100 keys. Included each month: 25 active tenants, 1B open-weight and 100M frontier routed tokens. Private plan.

Metered

Beyond the included amount

$0.25 per 1M open-weight tokens, $0.60 per 1M frontier tokens, $15 per additional active tenant per month.

An active tenant is a key that routed at least one token in the billing period. Prices as listed on 2026-10-15; the Marketplace listing is authoritative.

Data handling, in one paragraph

The gateway never stores or caches prompts and responses. For each request it records the key ID, route, deployment name, data boundary and token counts. Inference runs on your own deployments, in the region or data zone you chose; with x-dca-data-boundary: us the gateway uses only the deployments you declared for the United States and refuses a request it cannot serve inside the boundary instead of rerouting it. Retention at the model layer follows Microsoft's and the model publisher's terms for your deployment. Account and subscription data are stored in the United States. Details: security and privacy and our privacy notice.

Support: support@dcassociatesgroup.com · usage and support · documentation and samples