All new OpenAI resource deployment I created turns out to be "FAILED"

Question

I have successfully applied for the "access to OpenAI" to my company account a year ago, and I created a few OpenAI deployments successfully. I have been using them smoothly over these months.

Today, I planned to create a few more OpenAI deployment in different regions because I want to try more OpenAI models in different regions. I created four deployments today of different regions.

Unfortunately, all deployments today take over two hours each (as "Creating" status) and it turns out to be in "Failed" status with no reason.

Any clue of why it fails?

Answer 1

Based on the screenshot that you've posted here, the model that has failed is the one in "South India" location. It's likely that you've exceeded the token limits defined as part of the Azure quota. This is pretty common when you're running a deployment script multiple times a day or sharing an OpenAI subscription with other processes and thus leading to quota exhaustion. Try to re-create/re-deploy after some time and see if the deployment succeeds as there could be some intermittent capacity issues, transient faults etc. Also ensure that you're operating within the limits/quotas for the service defined here -- https://learn.microsoft.com/en-us/azure/ai-services/openai/quotas-limits

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.