Azure OpenAI (gateway)

To route Azure OpenAI through the Respan gateway, switch from the AzureOpenAI client to the standard OpenAI client. It is a small, mechanical change: the request and response shapes are identical (both classes live in the same openai package), so only the constructor changes.

Migrate from AzureOpenAI to OpenAI

$
# Python
$
pip install openai
$
$
# TypeScript
$
npm install openai

$
export RESPAN_API_KEY="YOUR_RESPAN_API_KEY"
$
export RESPAN_BASE_URL="https://api.respan.ai/api"
$
export RESPAN_MODEL="YOUR_MODEL_OR_DEPLOYMENT_SLUG"

1
from openai import AzureOpenAI
2
3
client = AzureOpenAI(
4
    azure_endpoint="https://YOUR_RESOURCE.openai.azure.com",
5
    api_key="YOUR_AZURE_OPENAI_API_KEY",
6
    api_version="2024-10-21",
7
)
8
9
response = client.chat.completions.create(
10
    model="your-azure-deployment",
11
    messages=[{"role": "user", "content": "Say hello in one sentence."}],
12
)

1
import { AzureOpenAI } from "openai";
2
3
const client = new AzureOpenAI({
4
  endpoint: "https://YOUR_RESOURCE.openai.azure.com",
5
  apiKey: process.env.AZURE_OPENAI_API_KEY,
6
  apiVersion: "2024-10-21",
7
});

What changes

Nothing else in your call sites changes.

Fallback to another provider

The gateway can fail over automatically. This is the common Azure-primary, Claude-fallback setup:

1
response = client.chat.completions.create(
2
    model=os.environ["RESPAN_MODEL"],
3
    messages=[{"role": "user", "content": "Say hello in one sentence."}],
4
    extra_body={"fallback_models": ["claude-sonnet-4-5-20250929"]},
5
)

Switch models

Change model to another Azure deployment or to any other model configured for your Respan gateway account.

1
client.chat.completions.create(model="YOUR_MODEL_OR_DEPLOYMENT_SLUG", messages=messages)
2
client.chat.completions.create(model="gpt-4o-mini", messages=messages)
3
client.chat.completions.create(model="claude-sonnet-4-5-20250929", messages=messages)

See the full model list.

Escape hatch: keep the AzureOpenAI client

Subclass the Azure client to skip its /deployments/{model} path rewrite, and send auth as a bearer token (the router does not accept the Azure api-key header):

1
import os
2
from openai import AzureOpenAI, OpenAI
3
4
class RespanAzureOpenAI(AzureOpenAI):
5
    def _build_request(self, options, *, retries_taken=0):
6
        return OpenAI._build_request(self, options, retries_taken=retries_taken)
7
8
respan_api_key = os.environ["RESPAN_API_KEY"]
9
10
client = RespanAzureOpenAI(
11
    base_url=os.environ["RESPAN_BASE_URL"],
12
    api_key=respan_api_key,
13
    api_version="2024-10-21",  # required by the client; ignored by the gateway
14
    default_headers={"Authorization": f"Bearer {respan_api_key}"},
15
)