Azure OpenAI – 0 quota in DataZoneStandard (Sweden Central) after approval on new tenant

Question

Hello,

I have a newly created Azure tenant and subscription.

The Azure OpenAI Limited Access application has been approved.

Region: Sweden Central

Deployment mode required: DataZoneStandard (EU data processing requirement)

However, all OpenAI models show 0 quota in this deployment mode.

In a previous subscription (different tenant), quota was automatically provisioned upon approval. I would like to confirm whether the current behavior (0 initial quota in DataZoneStandard for a new tenant) is expected.

Is a separate quota increase request required per model and per deployment mode in this scenario?

Thank you for your clarification.

Answer 1

For Azure OpenAI, quota behavior is subscription- and region-specific, and is defined per model and deployment type in each region.

From the provided information:

1. Quota is always scoped at the subscription level, per region, and per model/deployment type. It is not guaranteed that a new subscription will receive non‑zero default quota for every deployment mode or model in a region.

   “Tokens per minute (TPM) and requests per minute (RPM) limits are defined per region, per subscription, and per model or deployment type.”

2. Default quota is not uniform and is not retroactively or automatically increased in all scenarios. For provisioned/managed offerings, documentation explicitly states that:

   “New and existing subscriptions are assigned a small amount of provisioned quota in many regions… For existing customers, if the region already contains a quota assignment, the quota limit isn't changed for the region.”

   This shows that default quota can vary and that having 0 quota for a given configuration in a new subscription is an expected possibility.
3. When quota is 0 for a given model/deployment type/region, the supported path is to request quota via the quota experience:
   • In Azure AI Foundry portal: Management → Quota to view current allocations and usage.
   • Use the Request Quota link/icon next to the relevant quota item to submit a quota increase request.
4. For provisioned/managed quota, the self‑service model is explicitly per quota item (which aggregates deployments), and requests are made per quota type:

   “Customers no longer obtain quota by contacting their sales teams. Instead, they use the self-service quota request form and specify the PTU-Managed quota type.”

Applied to the described scenario (new tenant, Sweden Central, DataZoneStandard):

• Having 0 initial quota for DataZoneStandard in a new subscription is consistent with the documented behavior and can be expected.
• A separate quota request is required for the relevant quota item(s) in that region. In practice, this is done per quota type (for example, per model/deployment type or per model‑independent quota item, depending on how it is exposed in the Quota blade), not per individual deployment instance.
• After quota is granted for the DataZoneStandard deployment type in Sweden Central, deployments can be created within that quota.

Recommended actions:

1. In Azure AI Foundry portal, go to Management → Quota for the Sweden Central region and locate the DataZoneStandard/OpenAI quota entries.
2. Use the Request Quota action next to the relevant quota item(s) to request the needed TPM/RPM (or PTU/throughput) for the required models.

---

References:

• Azure OpenAI in Microsoft Foundry Models quotas and limits
• Manage Azure OpenAI in Azure AI Foundry Models quota
• Azure OpenAI provisioned Managed offering updates
• Azure OpenAI Provisioned Managed Offering in Azure Government