Share via

Unable to edit quota allocation for claude-opus-4-7 (East US 2) — generic error

Farrukh Umarov 0 Reputation points
2026-05-19T01:25:45.07+00:00

I'm unable to edit the quota allocation for our Claude model deployments in the Foundry portal. When I try to save a new allocation, I get a generic "An error occurred. Please reach out to support for additional assistance." message with no further detail.

Context:

  • Region: East US 2
  • Deployment: claude-opus-4-7 (Claude Opus 4.7, Global Standard)
  • Current allocation: 40K TPM
  • Attempted new allocation: 1801K TPM
  • Shared Claude pool: 2M TPM total (approved quota increase)
  • Other deployments in the shared pool: claude-haiku-4-5 (79K TPM), claude-sonnet-4-6 (80K TPM)

The 2M TPM increase for Claude models was approved more than 12 hours ago, and the right panel of the Quota page confirms "Total shared quota: 2M TPM" with "Remaining: 1.96M TPM" for this deployment. Despite that, every save attempt fails with the same generic error — including values well below the remaining headroom.

Foundry Models
Foundry Models

A catalog of AI models in Microsoft Foundry that you can discover, compare, and deploy using Azure’s built‑in tools for evaluation, fine‑tuning, and inference

3 answers

Sort by: Most helpful
Most helpful Newest Oldest
  1. kagiyama yutaka 3,510 Reputation points
    2026-05-22T05:52:42.7733333+00:00

    I think the generic quota error just means the portal can’t commit the quota update. if the Claude shared‑pool entitlement is out of sync, no value saves — even tiny ones. the fix is Azure Support triggering an Anthropic entitlement re‑sync.

    Was this answer helpful?

  2. Anshika Varshney 12,775 Reputation points Microsoft External Staff Moderator
    2026-05-19T02:41:10.7233333+00:00

    Hi Farrukh Umarov,

    Tthanks for sharing the details. I understand this can be confusing.

    From what you described, this behavior is actually expected in many cases for Claude models in Azure AI Foundry.

    Here is what is likely happening:

    Claude models like Opus use a different quota system compared to normal Azure OpenAI models. Their quota is not always available by default and is managed separately at the subscription level. [msnugget.com]

    If your quota shows as zero or if the edit option is disabled, it usually means that no quota has been assigned yet to your subscription. In this case, the portal does not allow you to edit or allocate anything, because there is nothing available to distribute.

    Also, for partner models like Claude, quota is not automatically created when you accept terms or deploy the model. Many subscriptions start with zero allocation and require backend enablement before quota can be used. [msnugget.com]

    This is why:

    The quota field can appear locked or not editable

    You cannot assign capacity to deployments

    Even after correct setup, it still does not work

    A few things you can double check:

    Go to quota section and confirm if it shows 0 out of 0 If yes, then this is not a configuration issue

    Make sure you are checking the correct subscription and region Quota is tied to subscription and region, not project

    Confirm Marketplace terms are accepted You already mentioned this, so that part looks fine

    If everything above is already verified, then your setup is correct and the behavior is due to quota not being provisioned yet on the backend.

    In simple words, the portal is not allowing changes because there is no quota assigned to your subscription for this model.

    Hope this helps clarify the situation. Do let me know if you have any further queries.

    Thankyou!

    Was this answer helpful?

    0 comments

  3. AI answer

    2026-05-19T01:26:00.5+00:00

    The behavior matches a quota-management failure in the Foundry portal rather than a model- or region-specific issue. From the available information, only the following actions are supported by the documented guidance:

    1. Confirm correct roles and subscription
    • Ensure the account has the roles required to edit quota allocations:
      • Cognitive Services Usages Reader at the subscription level.
      • Cognitive Services Contributor at the subscription level (combined with Usages Reader) to edit allocations.
    • Also verify Owner or Contributor on the subscription if a quota increase request is being applied or adjusted.
    1. Re-check quota view and propagation
    • In the Foundry portal, with New Foundry enabled:
      1. Go to Operate → Quota.
      2. Confirm the correct subscription and project are selected.
      3. Select the Claude deployment (for example, claude-opus-4-7) and verify in the details pane:
        • The model, region (East US 2), and deployment type are correct.
        • The Total shared quota and Remaining values reflect the 2M TPM pool and recent increase.
    • Quota changes can take up to 15 minutes to propagate. Since more than 12 hours have passed, the increase itself should be active; however, refreshing the Quota page after a short wait is still recommended when testing changes.
    1. Validate that the requested allocation is within model/region max TPM
    • Each model class has a max TPM per region that caps how much can be allocated to that model type, regardless of the shared pool size.
    • If the requested 1,801K TPM exceeds the model-class max TPM for claude-opus-4-7 in East US 2, the portal will not allow the allocation even if the shared pool shows sufficient remaining quota.
    • Try setting a significantly smaller allocation (for example, close to the current 40K TPM but higher) and gradually increase until reaching the highest value that saves successfully. If the error persists even for small increases, it indicates a portal or backend issue rather than a limit.
    1. Check generic quota troubleshooting conditions
    • If the Request quota button is disabled or the quota page behaves unexpectedly, confirm:
      • The correct subscription is selected.
      • The model and region combination is supported.
    • If quota changes are not reflected after approval and more than 24 hours have passed, the documented guidance is to contact Azure support.
    1. When the portal continues to return a generic error
    • The generic "An error occurred. Please reach out to support for additional assistance." message, combined with:
      • Valid roles,
      • Confirmed shared quota (2M TPM, with ~1.96M TPM remaining), and
      • Attempts with allocations well below remaining headroom, aligns with the documented case where quota changes fail and self-service troubleshooting is exhausted.
    • In this situation, the next step is to open a support ticket with Azure support so the backend quota configuration for the Claude model class in East US 2 can be checked and corrected.

    Summary of recommended steps:

    1. Verify roles: Cognitive Services Usages Reader + Cognitive Services Contributor, and Owner/Contributor on the subscription.
    2. Confirm correct subscription, project, and region in Operate → Quota.
    3. Test a small increase (for example, 50K–100K TPM) to see if any change is accepted.
    4. If all valid allocations fail and more than 24 hours have passed since the quota increase approval, contact Azure support as per the quota troubleshooting guidance.

    References:

    AI-generated content may be incorrect. Read our transparency notes for more information.

    Was this answer helpful?

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.