How much time does it take to create an OpenAI Endpoint?

Question

How much time does it take to create an OpenAI Endpoint?

It is VMS 270

Hello

I created an Azure AI resource, endpoint, deployment, etc.

The end point is <<deploymentname>>.openai.azure.com

I get a 400 error (for exactly the same request I had sent for a different endpoint & deployment): everything is the same & am going nuts to find out why it works on one & doesn't on the other. We are using the new OpenAI Responses API

Then, i realized that there is no DNS entry for the deployment where it is not working whereas it exists for the other one

So, 2 questions:

How long does it take for the DNS entries to propagate? It is a subdomain on openai.azure.com
Can this be the reason for the 400 error?

Thanks!

Adharsh Santhanam 8,230 Reputation points Microsoft Employee Moderator

2025-07-23T10:58:59.9633333+00:00

Hello It is VMS, Azure DNS propagation for subdomains like <deploymentname>.openai.azure.com typically takes a few minutes to a few hours, but in some very rare cases, can take 24-48 hours. Having said that, Azure DNS is designed to be fast and reliable, so delays beyond a few hours are pretty rare. A DNS resolution failure can lead to a 400 error. Try checking the DNS resolution using tools like nslookup (or any other tool for that matter) and verify that the deployment status is reflecting as "Succeeded". If the issue persists beyond a few hours, try deleting and recreating the endpoint.

Please do not forget to "Accept the answer” and “up-vote” wherever the information provided helps you, this can be beneficial to other community members.
Pavankumar Purilla 11,655 Reputation points Moderator

2025-07-24T06:31:55.7066667+00:00

Hi It is VMS,
Did you get any chance to check the response. Thank you!
Pavankumar Purilla 11,655 Reputation points Moderator

2025-07-25T02:43:42.1133333+00:00

Hi It is VMS,
Just following up to see if you had a chance to review the above response. Thank you!

1 answer

Your answer

Adharsh Santhanam 8,230 Reputation points Microsoft Employee Moderator

2025-07-23T10:58:59.9633333+00:00

Hello It is VMS, Azure DNS propagation for subdomains like <deploymentname>.openai.azure.com typically takes a few minutes to a few hours, but in some very rare cases, can take 24-48 hours. Having said that, Azure DNS is designed to be fast and reliable, so delays beyond a few hours are pretty rare. A DNS resolution failure can lead to a 400 error. Try checking the DNS resolution using tools like nslookup (or any other tool for that matter) and verify that the deployment status is reflecting as "Succeeded". If the issue persists beyond a few hours, try deleting and recreating the endpoint.

Please do not forget to "Accept the answer” and “up-vote” wherever the information provided helps you, this can be beneficial to other community members.
Pavankumar Purilla 11,655 Reputation points Moderator

2025-07-24T06:31:55.7066667+00:00

Hi It is VMS,
Did you get any chance to check the response. Thank you!
Pavankumar Purilla 11,655 Reputation points Moderator

2025-07-25T02:43:42.1133333+00:00

Hi It is VMS,
Just following up to see if you had a chance to review the above response. Thank you!

Answer 1

Hi It is VMS,
When creating a new Azure OpenAI resource and deploying a model, the endpoint (formatted as https://<resource-name>.openai.azure.com) is generally available within a few minutes. However, in some cases, DNS propagation for the subdomain can take a short while—typically up to 10–15 minutes. If the DNS entry hasn’t fully propagated, it could potentially cause connectivity issues, though it usually results in DNS resolution errors rather than a 400 Bad Request. A 400 error typically indicates that the request was received by the server but was malformed or incorrect For example, due to an invalid deployment name, incorrect API version, or misconfigured headers (especially when using the new OpenAI Responses API). We recommend waiting a few minutes and retrying the request. In parallel, please ensure that the deployment name is correct, the request format matches the expected schema for the Responses API, and the API key being used is associated with the correct resource.