Initial Global Standard quota allocation for claude-sonnet-5 in Sweden Central on a pay-as-you-go subscription

Thanks both. Resolved, and posting what actually worked in case it helps others.

The self-service form does accept an initial allocation from zero for Anthropic models, contrary to what I assumed when I asked. The form is Microsoft Foundry Service: Request for Quota Increase at https://aka.ms/oai/stuquotarequest and it is the right route even when the current quota is 0. Select Model Type Anthropic, then the hosting option, and the model list is filtered accordingly. Choosing Hosted on Azure lists the .Azure variants (claude-sonnet-5.Azure and so on), which is the bucket to target if you need in-region inference rather than inference on Anthropic infrastructure. The quota field is in capacity units where 1 unit is 1000 tokens per minute, so 450 means 450,000. No region is asked for on Global Standard, because that tier pools quota across all regions in the subscription.

I also opened an Azure support request first. That was unnecessary: the support engineer replied that quota approvals sit with a separate gating team and pointed me back to the same form. Worth skipping straight to the form.

One thing I could not confirm: whether pay-as-you-go is eligible for an initial Sonnet 5 allocation at all, which was my original question. The subscription already runs several Anthropic models including a newer one than Sonnet 5, so eligibility does not appear to be the blocker, but I will update this once the request is answered.

Thanks both. Resolved, and posting what actually worked in case it helps others.

The self-service form does accept an initial allocation from zero for Anthropic models, contrary to what I assumed when I asked. The form is Microsoft Foundry Service: Request for Quota Increase at https://aka.ms/oai/stuquotarequest and it is the right route even when the current quota is 0. Select Model Type Anthropic, then the hosting option, and the model list is filtered accordingly. Choosing Hosted on Azure lists the .Azure variants (claude-sonnet-5.Azure and so on), which is the bucket to target if you need in-region inference rather than inference on Anthropic infrastructure. The quota field is in capacity units where 1 unit is 1000 tokens per minute, so 450 means 450,000. No region is asked for on Global Standard, because that tier pools quota across all regions in the subscription.

I also opened an Azure support request first. That was unnecessary: the support engineer replied that quota approvals sit with a separate gating team and pointed me back to the same form. Worth skipping straight to the form.

One thing I could not confirm: whether pay-as-you-go is eligible for an initial Sonnet 5 allocation at all, which was my original question. The subscription already runs several Anthropic models including a newer one than Sonnet 5, so eligibility does not appear to be the blocker, but I will update this once the request is answered.

I think On a PAYG subscription, a Global Standard quota of 0 for claude‑sonnet‑5 means the initial quota hasn’t been assigned yet. Since initial allocations aren’t created automatically for every new Anthropic model, the path is to request an initial Global Standard quota through Microsoft support, after checkin in the claude model availability list that Sonnet‑5 GS is offered.

Welcome to Microsoft Q&A,

Hello @Artem Koloskov ,

In your case, I would recommend:

1. Open Microsoft Foundry > Operate > Quota and confirm that claude-sonnet-5 / Global Standard is shown with 0 allocation.
2. Submit a quota request for Claude Sonnet 5 for Sweden Central, specifying the required TPM/RPM and that the current allocation is zero.
3. If the request cannot be submitted because the form only allows increases from an existing allocation, or if the request is rejected, this is a good indication that the subscription requires a backend quota/allocation review.

Since you have already confirmed that the model is listed as available and your subscription is actively using other Anthropic models, I would include those details in the quota request.

If the portal does not provide a way to request an initial allocation from 0, I would recommend opening a Microsoft support request and asking specifically for an initial Claude Sonnet 5 Global Standard quota allocation, rather than requesting a generic model deployment issue.

Microsoft Foundry Models quotas and limits

Claude models in Microsoft Foundry

Microsoft documentation: https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/claude-models

If my answer helped you, please consider marking it as accepted. This helps others in the community find similar solutions.