fix(azure): support v1/preview/latest API versions

[Chesars]

• I understand that this repository is auto-generated and my pull request may not be merged

Changes being requested

Add support for Azure OpenAI v1/preview/latest API versions.

Currently, when using api_version="v1", "preview", or "latest", the SDK generates incorrect URLs:

 from openai import AzureOpenAI

 client = AzureOpenAI(
     api_version="preview",  # or "v1" or "latest"
     azure_endpoint="https://example-endpoint.openai.azure.com",
 )

# Both work with the fix:
 completion = client.chat.completions.create(
     model="deployment-name",
     messages=[{"role": "user", "content": "Hello!"}],
 )

 response = client.responses.create(
     model="deployment-name",
     input="Hello!",
 )
 
print(response)
print(completion)

	URL Generated
❌ Before	/openai/deployments/gpt-4o/chat/completions?api-version=preview
✅ After	/openai/v1/chat/completions?api-version=preview

The new Azure OpenAI v1 API uses /openai/v1/ path and the model is passed in the request body, not in the URL.

Non breaking changes - the values "v1", "preview", and "latest" previously resulted in 404 errors, so no existing working code is affected.

Changes:

• Add _is_v1_api flag to BaseAzureClient
• Skip adding /deployments/{model}/ path for v1 API in _build_request
• Use /openai/v1/ base URL for v1/preview/latest api_version values
• Add comprehensive tests (23 new test cases)

Note: The ?api-version= query parameter is still included for compatibility, as shown in the
https://learn.microsoft.com/en-us/azure/ai-foundry/openai/reference-preview-latest.

Additional context & links

• Fixes: AzureOpenAi is unable to process chat.completions when the API version is set to preview. #2584
• Azure docs: https://learn.microsoft.com/en-us/azure/ai-services/openai/api-version-deprecation

[chatgpt-codex-connector]

💡 Codex Review

openai-python/src/openai/lib/azure.py

Lines 79 to 83 in 2c95dde

	if self._azure_deployment and self._azure_endpoint and url not in _deployments_endpoints:
	merge_url = httpx.URL(url)
	if merge_url.is_relative_url:
	merge_raw_path = (
	self._azure_endpoint.raw_path.rstrip(b"/") + b"/openai/" + merge_url.raw_path.lstrip(b"/")

Keep v1 URLs when azure_deployment is supplied

For v1/preview/latest clients that still pass azure_deployment, _prepare_url rewrites any non-deployment endpoint to azure_endpoint/openai/<path> (lines 79-83) which drops the /v1 prefix set in the base URL. A call like AzureOpenAI(api_version="v1", azure_deployment="foo", ...).responses.create(...) will therefore hit /openai/responses?api-version=v1 and 404 instead of /openai/v1/responses. Because the constructor continues to store azure_deployment, this regression triggers whenever users reuse their old deployment parameter with the new v1 API.

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".