Repository navigation
Inquiry on Compatibility of Autogen Clients with Azure OpenAI v1 API #7040
Replies: 4 comments
|
Short answers based on current Autogen AgentChat / 1. Use
|
|
Short answers to your three questions, then the mechanism that justifies them. 1. Yes — Why: both clients are the same
|
|
Adding the one detail that decides between your two clients, since it's the actual failure mode people hit when they drop
from autogen_ext.models.openai import OpenAIChatCompletionClient
client = OpenAIChatCompletionClient(
model="<deployment-name>",
base_url="https://<resource>.openai.azure.com/openai/v1",
api_key=<key>,
model_info={"vision": False, "function_calling": True, "json_output": True, "family": "unknown"},
)On (3), the |
|
kuma5rou (@kuma5rou) — the decisive detail is that If that answers the compatibility question, please mark the comment above as the answer — the If you hit a different error on the v1 path, post it and I'll trace which client assumption it's tripping. |
Uh oh!
There was an error while loading. Please reload this page.
Autogen provides the OpenAIChatCompletionClient and AzureOpenAIChatCompletionClient classes, which are used to access the Azure OpenAI Service.
I'm currently considering switching to the next-generation v1 API (e.g., /v1/chat/completions, the OpenAI-compatible API) that your company provides. I would like to confirm the following points:
Questions for Confirmation
Is OpenAIChatCompletionClient fully compatible with the latest next-generation v1 API?
If I use the v1 API endpoint (e.g., /v1/chat/completions) instead of the Azure OpenAI–specific api-version approach, will it still be officially supported without issues?
If there are any related official documents or guidelines, could you kindly provide them as well?
I appreciate your time and assistance in confirming the above.
All reactions