Code examples

Start with a request. Build in the language you already use.

Before you run an example
Activate your subscription, connect a supported provider, and set DCA_GATEWAY_KEY in your server environment. Python also uses DCA_DEPLOYMENT for your deployment name and requires the openai package. Keep keys out of browser and mobile application code.

POST

bash

curl https://ai-gateway-east-us1.azure-api.net/gateway/openai/v1/chat/completions \
  -H "api-key: $DCA_GATEWAY_KEY" \
  -H "Content-Type: application/json" \
  -d '{"messages":[{"role":"user","content":"Hello, DCAI."}]}'

python

import os
from openai import OpenAI

key = os.environ["DCA_GATEWAY_KEY"]
client = OpenAI(
    base_url="https://ai-gateway-east-us1.azure-api.net/gateway/openai/v1",
    api_key=key,
    default_headers={"api-key": key},
)
response = client.chat.completions.create(
    model=os.environ["DCA_DEPLOYMENT"],
    messages=[{"role": "user", "content": "Hello, DCAI."}],
)
print(response.choices[0].message.content)

javascript

// Run on your server (Node.js), with the key in your environment.
const response = await fetch(
  "https://ai-gateway-east-us1.azure-api.net/gateway/openai/v1/chat/completions",
  {
    method: "POST",
    headers: {
      "api-key": process.env.DCA_GATEWAY_KEY,
      "Content-Type": "application/json",
    },
    body: JSON.stringify({
      messages: [{ role: "user", content: "Hello, DCAI." }],
    }),
  },
);
if (!response.ok) throw new Error(await response.text());
console.log(await response.json());

Your provider. Your deployment. One DCAI key.

Complete sample collection

The repository includes streaming, embeddings, Responses, Claude, route selection and connection checks. Python examples use the shared _gateway.py helper; download the complete sample set or clone the public repository.

python/chat.pyDownload ↓python/stream.pyDownload ↓python/embeddings.pyDownload ↓python/responses_api.pyDownload ↓python/claude.pyDownload ↓python/_gateway.pyDownload ↓python/requirements.txtDownload ↓javascript/chat.mjsDownload ↓javascript/claude.mjsDownload ↓javascript/package.jsonDownload ↓curl/chat.shDownload ↓curl/claude.shDownload ↓curl/routes.shDownload ↓

Repository setup instructions

Every sample reads your key from the environment and was run against the live gateway before publication.

bash

export DCA_GATEWAY_KEY="<your gateway key>"
export DCA_DEPLOYMENT="<a deployment in your Foundry, for example gpt-4.1-mini>"

Sample	What it shows
`python/chat.py`	Chat completion with the OpenAI SDK, and the gateway's response headers
`python/routes.py`	`x-dca-route`: the gateway picks your mapped deployment
`python/stream.py`	Streaming
`python/azure_openai_sdk.py`	Existing Azure OpenAI code, unchanged
`python/responses_api.py`	The Responses API
`python/embeddings.py`	Embeddings (`DCA_EMBEDDING_DEPLOYMENT`)
`python/claude.py`	Claude with the Anthropic SDK (`DCA_CLAUDE_DEPLOYMENT`)
`python/check_connection.py`	One request and a plain-language verdict, for troubleshooting
`javascript/chat.mjs`	Chat completion with the OpenAI Node.js SDK
`javascript/claude.mjs`	Claude with the Anthropic Node.js SDK
`curl/chat.sh`, `curl/routes.sh`, `curl/us-boundary.sh`, `curl/claude.sh`	The same with curl
`connect/connect-foundry.sh`	One-time setup in your tenant: the connector and its role

Python

bash

python3 -m venv .venv && . .venv/bin/activate
pip install -r samples/python/requirements.txt
python3 samples/python/chat.py

Node.js 18 or later

bash

cd samples/javascript && npm install && node chat.mjs