Azure OpenAI troubleshooting an error "No quota is available for this deployment. You can request for more quota."


Background Knowledge: Azure OpenAI Service Quota

Azure OpenAI Service is becoming increasingly popular. To provide a stable service to many users, it operates under a quota system. Users utilize the service within the allocated quota, and if they exceed this quota, they cannot deploy new AI models. The quota includes the number of OpenAI resources per Azure subscription region, tokens per minute, and the maximum number of deployed fine-tuned models. For frequent updates, please refer to the official documentation.

Inquiry Details

A customer wants to deploy a new AI model or edit an existing deployment on an existing Azure Open AI resource. However, even though they haven't exceeded the quota limit in Azure OpenAI, they receive an error in Azure Open AI Studio stating, "No quota is available for this deployment. You can request for more quota." They are unsure how to address this issue. The user has been assigned the "Cognitive Services Contributor" role for the "Open AI resource scope", and they believe they have sufficient permissions. They were able to deploy models in the past and are unsure about the cause of the current issue.

Response

Before June 2023, the "Cognitive Services Contributor" role alone was sufficient to deploy new models or edit existing deployments on Azure Open AI resources. However, around July 2023, the required permissions changed. For resources created after July 2023, assigning the "Cognitive Services Usages Reader role" scoped to the "Open AI resource deployment subscription" might resolve the issue. If this doesn't resolve the problem, it might be a bug, and it's recommended to contact Azure's support.

Conclusion

It's puzzling that an error message about quota appears when the actual issue is insufficient permissions. However, if you're aware of this, the solution is straightforward. There might be cases where the error message doesn't match the actual cause, making troubleshooting challenging. Hopefully, this article helps those facing similar issues.

References

Comments