Urgent Assistance Required – GPT-6 Quota 0 TPM, 20+ Requests Rejected

shiva 0 Reputation points
2026-10-19T22:51:44.7566667+00:00

Hello Microsoft Azure Team,

I represent Hapkit, a company using Azure Pay-As-You-Go for AI-powered developer tools and coding assistants.

Our new subscription has 0 TPM quota for multiple GPT-6 models.

We have submitted more than 20 quota requests across different regions, but every request has been denied due to regional capacity restrictions.

We are requesting an initial 200,000 TPM allocation, but we are willing to start with 10,000–50,000 TPM.

Could a Microsoft engineer please:

Review our subscription eligibility for GPT-6 model access.

Confirm whether our requests are being denied due to capacity or subscription-level restrictions.

Escalate this issue to the Microsoft Foundry quota and capacity management team.

Advise whether any supported region can provide an initial non-zero quota.

We can provide our subscription ID and rejected application IDs privately.

Thank you.

Hapkit

Azure Machine Learning
0 comments

1 answer

Sort by: Most helpful
Most helpful Newest Oldest
  1. Rukshan edirisinghe 2,365 Reputation points
    2026-10-20T02:56:35.27+00:00

    Hi @shiva

    I can't escalate on Microsoft's behalf from here, but I can explain why 20+ requests keep bouncing and what tends to work better than submitting more.

    Microsoft's quotas and limits page for Azure OpenAI in Foundry says two things that matter for a new pay-as-you-go subscription:

    • Your starting tier is based on your current usage of that model and your relationship with Microsoft (for example, enterprise agreements).
    • Increase requests are processed in order, and priority goes to customers who actively use their existing quota. Requests that don't meet that can be denied.

    So a brand-new subscription with zero usage asking for 200K TPM on the newest models is close to the worst case, whichever region you pick. Here's what usually moves things:

    1. Open the Quota page in the Foundry portal and look for any model or deployment type where you already have a non-zero default. Often that's an earlier model family or the Global Standard deployment type.
    2. Start building usage there. Deploy what you can, run real traffic, and stay within the default. That usage history is what the quota process weighs.
    3. Ask for a small increase first, for example 10K TPM on one model, one region, Global Standard. Then step up once it's in use. A narrow, modest request is much easier to approve than many parallel large ones.
    4. For anything time-critical, talk to a Microsoft account team or partner. That's the channel for capacity conversations beyond the self-service form. Share the subscription ID and request IDs only there, not in this thread.

    When you check the Quota page, is there any model with a non-zero allocation, or is it zero across the board?

    If this answered your question, please consider accepting it as the answer. If even the small request gets declined, let me know and I'll be happy to keep helping.

    Reference:

    Was this answer helpful?

    0 comments

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.