Can't verify the chat model using Azure API key.

[doin4]

• Disable all other plugins besides Copilot (required)
• Screenshot of note + Copilot chat pane + dev console added (required)

Copilot version:

(Bug report without the above will be closed)

Describe how to reproduce
I can't add Chat Models using Azure API key. I'm pretty sure I have the right API key and URL end point, but below error occurred. Besides, I'm confused about the Model Name, Instance Name and Deployment Name. I don't know how to fill those names. Once I changed the Instance Name, then I only got aborted error with on error in API key or end point (sadly still with an error).

Now I know I need to fill the right Instance Name to get the right URL, and I changed it to the right name, but it still occurred the error of 401 unauthorized. Again, I'm pretty sure the key and URL is correct. Appreciate it if anyone can give me any advise.

After this fail, I tried to add an embedding model also hosted by Azure, and I don't know why it worked.

UPDATE: I try to write a python script myself, and find that I can use the very API key and endpoint to access Azure model, don't know why it doesn't work on Obsidian.

Expected behavior
I want to verify my chat model using Azure API key.

Screenshots

Additional context
Add any other context about the problem here.

[doin4]

@logancyang Hi, thanks for this plug-in, and perhaps I know why this doesn't work. Here is the thing:

Abnormal

I checked the dev console and find that the url is re-directed to the place of "https://deepseek-r1.xxx" while the real url is "https://aaipxxxxx.azure.comxxx"

Try to find the right url

To verify the real url, I checked the dev console of cherry studio, and find the request url is "https://aaipxxxxx.azure.comxxx" and I can get the right response.

Furthermore, I use LLM to generate a python script to access the model and I can get the respons useing "https://aaipxxxxx.azure.comxxx".

Thus I'm wondering if there is an error in direct to the right url.

from openai import AzureOpenAI

# 设置环境变量，或者直接在代码中指定（不建议在生产环境中直接写死密钥）
os.environ["AZURE_OPENAI_API_KEY"] = "https://aaixxx.openai.azure.comxxxx"
os.environ["AZURE_OPENAI_ENDPOINT"] = "your_endpoint_url"

# 创建Azure OpenAI客户端
client = AzureOpenAI(
    api_key=os.getenv("AZURE_OPENAI_API_KEY"),
    api_version="2024-02-01",  # 根据实际使用的API版本进行调整
    azure_endpoint=os.getenv("AZURE_OPENAI_ENDPOINT")
)

# 使用客户端进行API调用，例如生成文本
response = client.chat.completions.create(
    model="your_deployment_name",  # 替换为你的模型部署名称
    messages=[
        {"role": "user", "content": "Hello, how are you?"}
    ]
)

print(response.choices[0].message.content)

In conclusion, I believe the issue lies within the URL component, specifically the current URL for Azure OpenAI. I would greatly appreciate it if you could address and resolve this bug.

[tacticsiege]

You appear to be attempting to add DeepSeek-R1, perhaps hosted in Azure AI Foundry.

Azure OpenAI is a different service (different Provider in terms of copilot settings) and DeepSeek isn't an OpenAI model.

You should be able to add your DeepSeek-R1 model as an "OpenAI Format" for the Provider, look at your DeepSeek-R1 deployment in Azure for using with OpenAI format chat completions.