Azure AI and Elevenlabs

the 'model id' that elevenlabs is asking for is actually your deployment name in azure ai. when you deployed your gpt model in azure, you gave it a deployment name. that is the name you need to use in the url and for the model id field. its not a separate hidden id.

so your url should look exactly like this: https://{your-azure-resource-name}.openai.azure.com/openai/deployments/{your-deployment-name}

and you would use that same {your-deployment-name} as the 'model id' in elevenlabs.

for the api version, use the same one that worked in postman. the one you mentioned, 2025-01-01-preview, is a good one to stick with.

the api key is found in the azure ai studio under 'keys and endpoints'. make sure you are using one of the listed keys.

if you are still getting an error, the most common culprit is the api version format. sometimes services are picky about the exact syntax. try it without the colon, like this in the elevenlabs field: api-version=2025-01-01-preview

also, check your azure ai resource's network settings. if you have any firewall rules enabled, it might be blocking the request from elevenlabs' servers. you might need to allow their ip addresses.

this might help in other api integrations too.

here is the azure docs on getting your key and endpoint. https://free.blessedness.top/azure/ai-services/openai/reference

hope this gets your voices talking.

Best regards,

Alex

and "yes" if you would follow me at Q&A - personaly thx.
P.S. If my answer help to you, please Accept my answer

https://ctrlaltdel.blog/

Hello Kaja,

Welcome to Microsoft Q&A and thank you for reaching out.

In Azure OpenAI, there isn’t a separate raw “model ID” like in the public OpenAI API. Instead, the deployment name you assigned when creating your model in Azure AI Studio acts as the “Model ID.” This is what you should provide in ElevenLabs wherever it asks for a model identifier.

If you’re looking for the base model reference (e.g., gpt-4.1, gpt-4o-mini, gpt-35-turbo), you can confirm this in the Model Catalog in Azure AI Studio. But for API calls, it’s always the deployment name that matters.

Your endpoint URL should follow this format:

https://{your-resource-name}.openai.azure.com/openai/deployments/{your-deployment-name}/chat/completions?api-version=2025-01-01-preview

{your-resource-name} = your Azure AI resource name.

{your-deployment-name} = the deployment name you created.

Don’t forget to include /chat/completions and append the api-version query parameter.

This is the exact URL you’ll need to configure in ElevenLabs.

You already tested successfully in Postman with:

?api-version=2025-01-01-preview

That’s the correct version. ElevenLabs must also include this parameter. Some integrations have a dedicated field for versioning, while others require you to append it directly to the endpoint URL.

You can find your API keys in the Azure Portal:

Go to your Azure OpenAI resource.

Under Keys and Endpoints, you’ll see Key1 and Key2.

Use either key and ensure it’s passed correctly in the header:

api-key: {your-key}
Content-Type: application/json

Postman succeeds because you explicitly set both the api-version parameter and the api-key header. ElevenLabs may be:

• Not appending the api-version correctly.
• Not sending the API key header in the right format.
• Or mistakenly expecting a raw “model ID” instead of your Azure deployment name.

To resolve the issue, please confirm:

Server URL matches the format shown above (with /chat/completions and the correct api-version).

Model ID in ElevenLabs is set to your Azure deployment name.

API Key is one of the valid keys from your Azure resource.

Headers include:

api-key: {your-key}
Content-Type: application/json

Use your deployment name as the Model ID.

Ensure your endpoint includes /chat/completions and ?api-version=2025-01-01-preview.

Double-check that ElevenLabs is passing your Azure API key in the correct header format.at.

I Hope this helps. Do let me know if you have any further queries.

If this answers your query, please do click Accept Answer and Yes for was this answer helpful.

Thank you!