Azure OpenAI – TPM quota drops by factor 100 on new Partner Success Core Benefits subscription

Question

Azure OpenAI – TPM quota drops by factor 100 on new Partner Success Core Benefits subscription

Tim Schröder 20

I'm running an agentic application on Azure OpenAI and recently had to switch subscriptions due to the new Partner benefits redemption process.

Context:

My previous subscription was based on a legacy yearly MPN sponsorship (MS-AZR-0025P) and had been running without any rate-limit issues for months.

After the latest renewal, I was forced to redeem my "Partner Success Core Benefits" through a brand-new subscription.

The new subscription is still in its early lifecycle and – as far as I understand – subject to dynamic/temporary quota throttling for new tenants or newly redeemed benefit subscriptions.

What's happening:

I deploy Azure OpenAI resources (models: GPT-5.x region: e.g. Sweden Central, US East 2, Poland ). Immediately after creation the provisioned throughput capacity (PTU) or tokens-per-minute (TPM) looks normal. However, a few hours after deployment, the TPM is silently reduced by a factor of ~100, e.g. from 1M TPM down to 10K TPM.

My agentic workloads fail with 429 Too Many Requests errors long before reaching any documented tier limit.

What I've already checked:

The Azure OpenAI quota blade shows the expected quota – the reduction is not reflected there. The behavior is reproducible across multiple model deployments and regions. There is no spending limit triggering resource deactivation (the subscription is active and has available credit). The same deployment configuration worked flawlessly on the previous (legacy) subscription.

My assumption: This looks like an internal rate-limiting policy applied specifically to new/fresh benefit subscriptions before they build up usage history. It doesn't match the documented default quotas for my region and feels like a dynamic, temporary enforcement layer.

Questions:

Is it a known behavior that newly redeemed "Partner Success Core Benefits" subscriptions receive reduced effective TPM quotas regardless of the provisioned model quota?

What is the recommended path – open a support ticket and request quota increase, or wait for an automated ramp-up?

Has anyone else encountered this and found a workaround (e.g. warm-up traffic, early quota request)?

Thanks in advance for any pointers!

Anshika Varshney 15,625 Zuverlässigkeitspunkte Externe Microsoft-Mitarbeiter Moderator

2026-06-03T10:12:21.16+00:00

Hello @Tim Schröder

Did you get any chance to review the response.

Thankyou!
Tim Schröder 20 Zuverlässigkeitspunkte

2026-06-03T11:49:54.6+00:00

Hello @Anshika Varshney

Thanks for your help. I will try to contact Partner Support first.
Anshika Varshney 15,625 Zuverlässigkeitspunkte Externe Microsoft-Mitarbeiter Moderator

2026-06-03T12:38:10.92+00:00

Hi @Tim Schröder

Please 'Upvote'(Thumbs-up) and 'Accept' as answer if the reply was helpful. This will be benefitting other community members who face the same issue.

Thankyou!

Antwort, die vom Frageautor angenommen wurde

0 zusätzliche Antworten

Ihre Antwort

Anshika Varshney 15,625 Zuverlässigkeitspunkte Externe Microsoft-Mitarbeiter Moderator

2026-06-03T10:12:21.16+00:00

Hello @Tim Schröder

Did you get any chance to review the response.

Thankyou!
Tim Schröder 20 Zuverlässigkeitspunkte

2026-06-03T11:49:54.6+00:00

Hello @Anshika Varshney

Thanks for your help. I will try to contact Partner Support first.
Anshika Varshney 15,625 Zuverlässigkeitspunkte Externe Microsoft-Mitarbeiter Moderator

2026-06-03T12:38:10.92+00:00

Hi @Tim Schröder

Please 'Upvote'(Thumbs-up) and 'Accept' as answer if the reply was helpful. This will be benefitting other community members who face the same issue.

Thankyou!

Answer 1

Hello @Tim Schröder

This is expected behavior when switching to a Partner Success Core Benefits subscription. The TPM quota drop is directly related to the subscription type and quota tier system introduced for Azure OpenAI.

Request a Quota Increase

Option 1: Request Quota Increase Through Azure Portal[learn.microsoft]

Go to Azure Portal → Azure OpenAI Resource

Navigate to Quota blade in the left menu

Select your region and model

Click Request Quota Increase

Enter desired TPM (e.g., 100,000 or higher)

Submit and wait for approval

Option 2: Use Quota Management Form[learn.microsoft]

Use the official Azure OpenAI quota request form

Mention you're a Partner Success Core Benefits customer

Specify your business need for higher TPM

Option 3: Contact Partner Support

Since you're on Partner Success Core Benefits:

Contact Microsoft Partner Support directly

They can manually adjust your quota tier based on your partner status

Mention your payment history and expected usage

Quotas Are Per-Region, Per-Model:[github]

Request TPM increases for each region where you need capacity

Each model (GPT-4, GPT-3.5-turbo, etc.) has separate quota

Automatic Tier Upgrades: [learn.microsoft]

As your usage grows, Foundry can automatically move you to higher tiers
Microsoft considers payment history for eligibility
Previously approved quota increases are retained

This is not a bug it's the default quota assignment for Partner Success Core Benefits subscriptions. Requesting a quota increase will resolve the issue. Let me know if you need help with the quota request process
Thankyou!