How can I increase the RPM quota for gpt-5.6-luna when it is missing from the Azure quota form?

Question

I need help increasing the RPM quota for an existing Azure OpenAI deployment using gpt-5.6-luna.

Deployment details

• Subscription: Azure subscription 1
• Resource group: rg-jagan
• Azure OpenAI resource: oai-s3mtech-southindia
• Deployment name: gpt-5.6-luna
• Region: South India
• Deployment type/SKU: Global Standard
• Model version: 2026-07-09
• Provisioning state: Succeeded

The current deployment limits are:

Request rate: 1,000 requests per 60 seconds

Token rate: 1,000,000 tokens per 60 seconds

Capacity: 1,000

I need to increase the RPM quota from 1,000 RPM to 5,000 RPM for an agentic coding platform. A single coding task can generate multiple model calls for planning, tool execution, tool-result continuations, corrections, and retries. We are hitting the current RPM limit multiple times during a single user operation.

I tried to use the Azure quota-increase workflow:

Request for Quota Increase → Foundry Models → Global Standard → South India

However:

1. The form requires a model selection.
2. gpt-5.6-luna is not available in the model dropdown.
3. There is no Other, Model not listed, or free-text option.
4. I cannot submit the form without selecting a model.
5. Selecting another model would submit a quota request for the wrong model.
6. Azure Support and Troubleshooting does not provide a un for this quota request and redirects me back to thequota-increase form.

Request for help

Could a Microsoft/Azure representative please advise:

1. What is the correct process for increasing the RPM quota for gpt-5.6-luna?
2. Is gpt-5.6-luna eligible for quota increases under Glo
3. Why is this model deployed successfully but missing from the quota-increase form?
4. Can this request be escalated manually, or can the mod
5. If an RPM increase is not currently supported, what is the recommended way to handle higher request volume for this model?

I will use application-side rate limiting, centralized queuing, concurrency controls, capped retries, exponential backoff with jitter, and

usage monitoring.

Thank you.