Can Azure for Students subscription apply for AI model quota with token-based billing?

An Azure for Students subscription can request Azure OpenAI quota, but its offer-specific limits are substantially lower and some models receive zero quota; pay-as-you-go is not mandatory to submit a request, although it may be necessary when the student offer cannot provide the required model or capacity.

Check and use the quota already assigned

Quota and token-based billing are separate:

• TPM quota controls the rate at which the deployment can process tokens.
• Token-based billing charges for actual input and output usage; it is not a way to purchase or guarantee more TPM quota. ^1^

Check the allocation for the exact model and deployment type:

1. Sign in to Microsoft Foundry and enable New Foundry.
2. Open the relevant project.
3. Select Manage → Quota → Token per minute.
4. Check whether the selected model has a nonzero allocation.
5. If quota is already available, allocate it to the deployment using the pencil icon. Allow up to 15 minutes for the change to propagate. ^1^

Viewing subscription quota requires Cognitive Services Usages Reader at the subscription scope. Assigning that role requires role-assignment permission; otherwise, ask the subscription owner or administrator to assign it.

Current Foundry quota management is transitioning to subscription-level pools. For applicable models, Global Standard deployments share a subscription-wide pool, while Data Zone Standard deployments share a pool within each data zone. Model and deployment-type behavior can still vary during the rollout. ^2^

Request additional quota

If the available allocation is insufficient:

1. In Manage → Quota, select Request quota.
2. Alternatively, submit the official quota increase request form.
3. Specify the subscription, model, deployment type, required capacity, and expected usage.
4. Continue using existing quota where possible. Microsoft prioritizes customers who actively consume their existing allocation, and a request can be denied. ^1^

Azure for Students has offer-specific restrictions. The published table lists 1K TPM for supported models, with 0 quota for o-series, GPT-4.1, and GPT-4.5 Preview under that offer. Treat these as model-specific offer ceilings, not a universal quota for every current Foundry deployment. ^2^

You can verify that the subscription is actually using the Azure for Students offer:

az login
az rest --method GET \
  --uri "https://management.azure.com/subscriptions/[Subscription ID]"

The expected student quotaId is AzureForStudents_2018-01-01. If Foundry or the capacity API displays a different allocation from the published offer table, the value returned for that specific subscription is the effective value; an account-specific discrepancy must be addressed through the quota request process or Microsoft support.

When pay-as-you-go is needed

Upgrading to pay-as-you-go is appropriate when:

• the required model has zero student quota;
• the quota request is denied under the student offer;
• available regions or deployment types cannot satisfy the workload; or
• the student credit has expired or been exhausted.

Azure for Students can be upgraded to pay-as-you-go in the Azure portal. Open Subscriptions, select the student subscription, and use the available Upgrade action to complete the billing setup. ^3^

An upgrade changes the subscription from the student offer and makes it subject to non-student quota policies, but it does not guarantee approval, regional capacity, or access to a particular model. Pay-as-you-go subscriptions also do not support Azure’s credit-based spending limit, so configure budgets and cost alerts instead. ^4^

Alternatives without upgrading

• Use a supported model that already shows nonzero quota in Manage → Quota.
• Reduce TPM assigned to another deployment and reallocate it to the required deployment.
• For a Microsoft Learn exercise, use its Launch sandbox, guided project, or pre-provisioned environment when that specific module provides one.
• For development that does not require Azure-hosted endpoints, run an appropriate open-source model locally; this avoids Azure quota and billing but does not reproduce Azure OpenAI service behavior.

References

1. Manage Azure OpenAI in Microsoft Foundry Models quota
2. Azure OpenAI in Azure AI Foundry Models quotas and limits
3. Upgrade your Azure free account or Azure for Students Starter account
4. Azure spending limit