InsufficientQuota

Question

InsufficientQuota

Mike-E-angelo 631

I am experiencing the following:

ERROR: {"code": "InvalidTemplateDeployment", "message": "The template deployment '...' is not valid according to the validation procedure. The following resource provider(s) - 'Microsoft.CognitiveServices/accounts (2025-12-01)' reported preflight validation errors. Tracking id is '121b3c51-93b5-4416-bed3-0bbe6b53fb6d'. See inner errors for details."}
Inner Errors: 
{"code": "InsufficientQuota", "message": "This operation require 10 new capacity in quota Requests Per Minute - GPT Image 1 Mini Generation, which is bigger than the current available capacity 0. The current quota usage is 0 and the quota limit is 0 for quota Requests Per Minute - GPT Image 1 Mini Generation."}

Usually when I encounter this, I have accounts that can be purged. But when I list:

> az cognitiveservices account list-deleted
[]

Nothing is displayed to delete.

How do I fix this? Thank you for any assistance you can provide.

Sai Kiran Mudavath 90 Reputation points Microsoft External Staff Moderator

2026-09-30T10:49:43.09+00:00
HI @Mike-E-angelo

Thanks for providing the complete error details.

The error confirms that the deployment is requesting 10 RPM for GPT Image 1 Mini Generation, while the currently available capacity and quota limit reported for that quota are both 0. The az cognitiveservices account list-deleted result is empty, so there are no deleted accounts to purge.

Since you also don't see a Request quota option in the portal, there are a couple of things we should verify before determining the next step:

Please confirm the Azure region where you are trying to create the deployment.

In the Azure portal, make sure you are viewing the correct subscription.

If the Quota page is empty, verify that your account has the Cognitive Services Usages Reader role at the subscription level. Microsoft documents that this role is required to view quota allocations.

If the Request quota option is unavailable or disabled, Microsoft notes that this can also depend on your subscription permissions and whether the particular model/region combination supports quota increases.

For the CLI output, you don't need to review the entire 157 KB response manually. The following command can display the quota entries in a more readable format:

az cognitiveservices usage list -l <region> -o table

The az cognitiveservices usage list command returns the quota usage for the selected region and subscription.

Please share the target region and, if possible, the relevant output for the GPT Image 1 Mini Generation quota. With those details, we can determine whether the quota needs to be requested, whether the model/region has a different quota-management path, or whether the issue requires further investigation.

Thank You!
Mike-E-angelo 631 Reputation points

2026-09-30T12:19:25.54+00:00

Hi @Sai Kiran Mudavath thank you for your reply it is appreciated. Please note that I have deleted all foundries:

And have verified that all subscriptions have no deleted resources after pressing Manage deleted resources:

Nonetheless the error persists when I attempt to create a foundry/model in any subscription now. I would greatly appreciate any priority you can provide as this is blocking us from deploying our application. Thank you again for your efforts.

Mike-E-angelo 631

Hi @Sai Kiran Mudavath pardon the previous reply I did not see your reply. Yes, in regards to region this is EastUS2. Here's the result of usage:

> az cognitiveservices usage list -l EastUS2 -o table 
CurrentValue    Limit      Unit
--------------  ---------  ------
0.0             0.0        Count
0.0             0.0        Count
0.0             10.0       Count
0.0             30.0       Count
0.0             3.0        Count
0.0             0.0        Count
0.0             300.0      Count
0.0             500.0      Count
0.0             500.0      Count
0.0             500.0      Count
0.0             1000000.0  Count
0.0             40.0       Count
0.0             80.0       Count
0.0             0.0        Count
0.0             250.0      Count
0.0             250.0      Count
0.0             250.0      Count
0.0             500.0      Count
0.0             500.0      Count
0.0             30000.0    Count
0.0             80000.0    Count
0.0             200000.0   Count
0.0             1000000.0  Count
0.0             0.0        Count
0.0             150.0      Count
0.0             450.0      Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             30000.0    Count
0.0             100000.0   Count
0.0             400.0      Count
0.0             400.0      Count
0.0             50.0       Count
0.0             50.0       Count
0.0             350.0      Count
0.0             0.0        Count
0.0             0.0        Count
0.0             350.0      Count
0.0             350.0      Count
0.0             1000.0     Count
0.0             0.0        Count
0.0             1000.0     Count
0.0             0.0        Count
0.0             50.0       Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             100000.0   Count
0.0             100000.0   Count
0.0             1000000.0  Count
0.0             0.0        Count
0.0             100000.0   Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             60.0       Count
0.0             60.0       Count
0.0             0.0        Count
0.0             100.0      Count
0.0             1000.0     Count
0.0             500.0      Count
0.0             500.0      Count
0.0             250.0      Count
0.0             500.0      Count
0.0             200000.0   Count
0.0             30000.0    Count
0.0             0.0        Count
0.0             0.0        Count
0.0             1000.0     Count
0.0             0.0        Count
0.0             250.0      Count
0.0             250.0      Count
0.0             500.0      Count
0.0             500.0      Count
0.0             500.0      Count
0.0             500.0      Count
0.0             500.0      Count
0.0             500.0      Count
0.0             200.0      Count
0.0             5000.0     Count
0.0             1000000.0  Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             200000.0   Count
0.0             30000.0    Count
0.0             200000.0   Count
0.0             100000.0   Count
0.0             30000.0    Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             500.0      Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             200000.0   Count
0.0             200000.0   Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             30.0       Count
0.0             40.0       Count
0.0             400.0      Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             200000.0   Count
0.0             200000.0   Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             200000.0   Count
0.0             200000.0   Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             200000.0   Count
0.0             200000.0   Count
0.0             1000.0     Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             100.0      Count
0.0             600.0      Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             250.0      Count
0.0             250.0      Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             100.0      Count
0.0             100.0      Count
0.0             5000.0     Count
0.0             5000.0     Count
0.0             1000.0     Count
0.0             30.0       Count
0.0             30.0       Count
0.0             30.0       Count
0.0             30.0       Count
0.0             60.0       Count
0.0             60.0       Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             300.0      Count
0.0             300.0      Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             300.0      Count
0.0             300.0      Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             100.0      Count
0.0             300.0      Count
0.0             300.0      Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             4.0        Count
0.0             4.0        Count
0.0             100.0      Count
0.0             60.0       Count
0.0             40.0       Count
0.0             4.0        Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             300.0      Count
0.0             300.0      Count
0.0             1000.0     Count
0.0             300.0      Count
0.0             1000.0     Count
0.0             300.0      Count
0.0             100.0      Count
0.0             250.0      Count
0.0             250.0      Count
0.0             1000.0     Count
0.0             300.0      Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             60.0       Count
0.0             40.0       Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             300.0      Count
0.0             100.0      Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             250.0      Count
0.0             1000.0     Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             1000.0     Count
0.0             1000.0     Count
0.0             0.0        Count
0.0             0.0        Count
0.0             0.0        Count
0.0             100.0      Count
0.0             100.0      Count
0.0             500.0      Count
0.0             200.0      Count

What is frustrating is that I have had a model deployed to my Development subscription for over a year now without any issue. I did not have to assign any permissions, everything "just worked." However, after deleting that foundry and attempting to recreate it, I run into the same problem again. So, something seems to have gone wrong in the deleting and recreating Foundries.

Mike-E-angelo 631 Reputation points

2026-09-30T12:36:22.7833333+00:00

Hi @Sai Kiran Mudavath pardon the chatter but I have been digging into this. I was incorrect about the Development Subscription. I was able to recreate the Model in Development subscription, but not the Preview subscription (where we are trying to deploy). So it would seem that quota has been assigned in Development, but not in Preview (and other subscriptions)? If that is the case, I need to find the way to ask for quota in these subscriptions for gpt-image-1.5, please? Thank you for any assistance. It's weird because I was able to deploy, delete, and redeploy just fine to Preview earlier this morning (numerous times).
Mike-E-angelo 631 Reputation points

2026-09-30T16:45:17.4733333+00:00

Hi @Sai Kiran Mudavath after some poking around I found the quota request form. Application IDs are #60071, #60075, and #60095. It looks like #60095 was already declined because EASTUS2 is not accepting new capacity. Why is it showing as an option on the request form if it's not available? This is really confusing. Are you able to assist? We cannot deploy our application until we resolve this. Thank you for your assistance.
Sai Kiran Mudavath 90 Reputation points Microsoft External Staff Moderator

2026-10-01T01:06:09.4333333+00:00
Hi @Mike-E-angelo ,

Thank you for the additional details. The new information helps narrow this down.

I can see that the az cognitiveservices usage command is returning quota information for EastUS2, but the table output does not include the quota names, so we cannot identify the specific GPT-image quota from those rows alone. Microsoft documents that the quota name is available through name.value.

Since you have multiple subscriptions, please run the following against the Preview subscription:

az cognitiveservices usage list -l EastUS2 --subscription "<Preview-subscription-id>" --query "[?contains(name.value, 'gpt-image')].{Quota:name.value,CurrentValue:currentValue,Limit:limit,Unit:unit}" -o table

This will return only the quota entries related to the GPT-image models instead of the full output.

Please redact any subscription ID, tenant ID, resource ID, or other identifying information from the output before posting it publicly. If needed, you can share such details with me through a private message.

I also noticed that the original deployment error refers to GPT Image 1 Mini, while your latest comment refers to gpt-image-1.5. These are separate models, so please confirm which model you are currently trying to deploy and the deployment type.

Regarding request #60095, the message that EastUS2 is not accepting additional capacity is also important. Microsoft provides a separate Model Capacities API to determine available deployment capacity for a specific model and region. Once we confirm the exact model and deployment type, we can distinguish a subscription-quota issue from a regional capacity issue.

Thank you.
Mike-E-angelo 631 Reputation points

2026-10-01T06:12:20.2366667+00:00
Hi @Sai Kiran Mudavath thank you for your continued assistance, and also for the simplified query. Here are the results in Preview:

Quota CurrentValue Limit Unit -------------------------------------------- -------------- ------- ------ OpenAI.GlobalStandard.gpt-image-1 0.0 0.0 Count OpenAI.GlobalStandard.gpt-image-1-mini 0.0 0.0 Count OpenAI.GlobalStandard.gpt-image-1.5 0.0 0.0 Count OpenAI.DataZoneStandard.gpt-image-1.5 0.0 0.0 Count OpenAI.GlobalStandard.gpt-image-2 0.0 0.0 Count OpenAI.GlobalStandard.gpt-image-2.5-flare 0.0 0.0 Count OpenAI.GlobalStandard.gpt-image-2.5-sunburst 0.0 0.0 Count

Also, pardon the confusion; as all three applications state, the model in question is gpt-image-1.5. Mini was a previous model under test and was part of our Bicep deployments, but all 3 applications provided yesterday (along with follow-ups #60095 and #60106) all ask for gpt-image-1.5.

For #06071, #60095, and #60106, they were all denied outright -- all for the Preview subscription, which we are deploying to. The other applications haven't received a reply yet, so I'm wondering if something is wrong with Preview. We have an email from Application #3347365 in January saying this subscription has been accepted, so this is really confusing to me:

Thank you for your continued assistance. It would be valuable to get this addressed quickly as it is currently preventing us from deploying our application to our desired subscriptions.
Sai Kiran Mudavath 90 Reputation points Microsoft External Staff Moderator

2026-10-01T09:40:32.78+00:00

Hi @Mike-E-angelo ,

Thank you for providing the additional details. The filtered results confirm that the Preview subscription currently has a quota limit of 0 for gpt-image-1.5 for both GlobalStandard and DataZoneStandard in EastUS2.

The earlier onboarding confirmation you received indicates that the subscription was approved for gpt-image-1.5 access. However, Microsoft documents model access and quota as separate requirements: gpt-image-1.5 is a limited-access model, and a deployment also requires an applicable quota allocation.

Microsoft also documents that quota-increase requests are evaluated individually and that requests may be denied. Therefore, the current 0 quota explains the InsufficientQuota error, but the public documentation does not provide a way to manually restore a denied quota allocation from the portal or CLI.

Since you have already submitted requests #60071, #60095, and #60106 and they were denied, the next step is to follow the decision provided for those quota requests. If the request decision indicates that additional quota is not currently available for the requested deployment, the documented alternatives are to use a deployment/region for which quota is available or submit a new quota request when eligible. Microsoft also provides the Model Capacities check to distinguish quota allocation from available deployment capacity.

Based on the information currently available, there is no additional local configuration change we can recommend to allocate the missing quota in the Preview subscription.

Thank you.
Mike-E-angelo 631 Reputation points

2026-10-01T10:41:00.4166667+00:00

Hi @Sai Kiran Mudavath thank you for your reply and for your patience as I understand what is going on with this really confusing situation. No other resource in Azure has been nearly as convoluted and involved as this. It's unclear why so much effort is being made to prevent users from actually using your resources.

So, if I understand correctly, my Preview environment was approved but given zero quota? I understand having to follow the decision, but that makes sense only if I actually have quota and can use the subscription in some capacity. As it stands now, my subscription has zero quota for the "approved" models and cannot be used at all.

This makes no sense to me. Why approve a model for use when no quota is given? It would seem at the very least, a minimum quota would be provided with an "approved" subscription. Otherwise, the subscription is worthless for any deployments.

I would greatly appreciate any confirmation of my understanding, as well as any corrections. Thank you for your continued assistance.
Sai Kiran Mudavath 90 Reputation points Microsoft External Staff Moderator

2026-10-02T01:16:33.47+00:00

Hi @Mike-E-angelo ,

Yes, your understanding of the current state is correct. Your subscription was previously approved for gpt-image-1.5 access, but the Preview subscription currently shows 0 quota for gpt-image-1.5 for both GlobalStandard and DataZoneStandard.

The important distinction is that model access and quota are separate parts of the deployment process. Microsoft documents gpt-image-1.5 as a limited-access preview model, while quota is managed separately for deployments.

Therefore, the previous onboarding confirmation establishes that the subscription was granted access to the model; it does not, by itself, establish that a non-zero deployment quota is currently allocated. In your case, the current 0 quota is what is causing the InsufficientQuota deployment failure.

Microsoft documents that quota increase requests are evaluated individually and may be denied. Since you have already submitted multiple quota requests and they were denied, there is no documented portal or CLI setting that can manually allocate that missing quota.

So, based on the information currently available, your interpretation is correct: the subscription can have model access while still having no deployment quota available for that model. The reason the quota was not allocated to the Preview subscription despite the earlier onboarding approval is not something the public Q&A documentation allows us to determine.

Thank you!

2 answers

Your answer

Sai Kiran Mudavath 90 Reputation points Microsoft External Staff Moderator

2026-09-30T10:49:43.09+00:00

HI @Mike-E-angelo

Thanks for providing the complete error details.

The error confirms that the deployment is requesting 10 RPM for GPT Image 1 Mini Generation, while the currently available capacity and quota limit reported for that quota are both 0. The az cognitiveservices account list-deleted result is empty, so there are no deleted accounts to purge.

Since you also don't see a Request quota option in the portal, there are a couple of things we should verify before determining the next step:

Please confirm the Azure region where you are trying to create the deployment.

In the Azure portal, make sure you are viewing the correct subscription.

If the Quota page is empty, verify that your account has the Cognitive Services Usages Reader role at the subscription level. Microsoft documents that this role is required to view quota allocations.

If the Request quota option is unavailable or disabled, Microsoft notes that this can also depend on your subscription permissions and whether the particular model/region combination supports quota increases.

For the CLI output, you don't need to review the entire 157 KB response manually. The following command can display the quota entries in a more readable format:

az cognitiveservices usage list -l <region> -o table

The az cognitiveservices usage list command returns the quota usage for the selected region and subscription.

Please share the target region and, if possible, the relevant output for the GPT Image 1 Mini Generation quota. With those details, we can determine whether the quota needs to be requested, whether the model/region has a different quota-management path, or whether the issue requires further investigation.

Thank You!
Mike-E-angelo 631 Reputation points

2026-09-30T12:19:25.54+00:00

Hi @Sai Kiran Mudavath thank you for your reply it is appreciated. Please note that I have deleted all foundries:

And have verified that all subscriptions have no deleted resources after pressing Manage deleted resources:

Nonetheless the error persists when I attempt to create a foundry/model in any subscription now. I would greatly appreciate any priority you can provide as this is blocking us from deploying our application. Thank you again for your efforts.
Mike-E-angelo 631 Reputation points

2026-09-30T12:36:22.7833333+00:00

Hi @Sai Kiran Mudavath pardon the chatter but I have been digging into this. I was incorrect about the Development Subscription. I was able to recreate the Model in Development subscription, but not the Preview subscription (where we are trying to deploy). So it would seem that quota has been assigned in Development, but not in Preview (and other subscriptions)? If that is the case, I need to find the way to ask for quota in these subscriptions for gpt-image-1.5, please? Thank you for any assistance. It's weird because I was able to deploy, delete, and redeploy just fine to Preview earlier this morning (numerous times).
Mike-E-angelo 631 Reputation points

2026-09-30T16:45:17.4733333+00:00

Hi @Sai Kiran Mudavath after some poking around I found the quota request form. Application IDs are #60071, #60075, and #60095. It looks like #60095 was already declined because EASTUS2 is not accepting new capacity. Why is it showing as an option on the request form if it's not available? This is really confusing. Are you able to assist? We cannot deploy our application until we resolve this. Thank you for your assistance.
Sai Kiran Mudavath 90 Reputation points Microsoft External Staff Moderator

2026-10-01T01:06:09.4333333+00:00

Hi @Mike-E-angelo ,

Thank you for the additional details. The new information helps narrow this down.

I can see that the az cognitiveservices usage command is returning quota information for EastUS2, but the table output does not include the quota names, so we cannot identify the specific GPT-image quota from those rows alone. Microsoft documents that the quota name is available through name.value.

Since you have multiple subscriptions, please run the following against the Preview subscription:

az cognitiveservices usage list -l EastUS2 --subscription "<Preview-subscription-id>" --query "[?contains(name.value, 'gpt-image')].{Quota:name.value,CurrentValue:currentValue,Limit:limit,Unit:unit}" -o table

This will return only the quota entries related to the GPT-image models instead of the full output.

Please redact any subscription ID, tenant ID, resource ID, or other identifying information from the output before posting it publicly. If needed, you can share such details with me through a private message.

I also noticed that the original deployment error refers to GPT Image 1 Mini, while your latest comment refers to gpt-image-1.5. These are separate models, so please confirm which model you are currently trying to deploy and the deployment type.

Regarding request #60095, the message that EastUS2 is not accepting additional capacity is also important. Microsoft provides a separate Model Capacities API to determine available deployment capacity for a specific model and region. Once we confirm the exact model and deployment type, we can distinguish a subscription-quota issue from a regional capacity issue.

Thank you.
Mike-E-angelo 631 Reputation points

2026-10-01T06:12:20.2366667+00:00

Hi @Sai Kiran Mudavath thank you for your continued assistance, and also for the simplified query. Here are the results in Preview:

Quota CurrentValue Limit Unit -------------------------------------------- -------------- ------- ------ OpenAI.GlobalStandard.gpt-image-1 0.0 0.0 Count OpenAI.GlobalStandard.gpt-image-1-mini 0.0 0.0 Count OpenAI.GlobalStandard.gpt-image-1.5 0.0 0.0 Count OpenAI.DataZoneStandard.gpt-image-1.5 0.0 0.0 Count OpenAI.GlobalStandard.gpt-image-2 0.0 0.0 Count OpenAI.GlobalStandard.gpt-image-2.5-flare 0.0 0.0 Count OpenAI.GlobalStandard.gpt-image-2.5-sunburst 0.0 0.0 Count

Also, pardon the confusion; as all three applications state, the model in question is gpt-image-1.5. Mini was a previous model under test and was part of our Bicep deployments, but all 3 applications provided yesterday (along with follow-ups #60095 and #60106) all ask for gpt-image-1.5.

For #06071, #60095, and #60106, they were all denied outright -- all for the Preview subscription, which we are deploying to. The other applications haven't received a reply yet, so I'm wondering if something is wrong with Preview. We have an email from Application #3347365 in January saying this subscription has been accepted, so this is really confusing to me:

Thank you for your continued assistance. It would be valuable to get this addressed quickly as it is currently preventing us from deploying our application to our desired subscriptions.
Sai Kiran Mudavath 90 Reputation points Microsoft External Staff Moderator

2026-10-01T09:40:32.78+00:00

Hi @Mike-E-angelo ,

Thank you for providing the additional details. The filtered results confirm that the Preview subscription currently has a quota limit of 0 for gpt-image-1.5 for both GlobalStandard and DataZoneStandard in EastUS2.

The earlier onboarding confirmation you received indicates that the subscription was approved for gpt-image-1.5 access. However, Microsoft documents model access and quota as separate requirements: gpt-image-1.5 is a limited-access model, and a deployment also requires an applicable quota allocation.

Microsoft also documents that quota-increase requests are evaluated individually and that requests may be denied. Therefore, the current 0 quota explains the InsufficientQuota error, but the public documentation does not provide a way to manually restore a denied quota allocation from the portal or CLI.

Since you have already submitted requests #60071, #60095, and #60106 and they were denied, the next step is to follow the decision provided for those quota requests. If the request decision indicates that additional quota is not currently available for the requested deployment, the documented alternatives are to use a deployment/region for which quota is available or submit a new quota request when eligible. Microsoft also provides the Model Capacities check to distinguish quota allocation from available deployment capacity.

Based on the information currently available, there is no additional local configuration change we can recommend to allocate the missing quota in the Preview subscription.

Thank you.
Mike-E-angelo 631 Reputation points

2026-10-01T10:41:00.4166667+00:00

Hi @Sai Kiran Mudavath thank you for your reply and for your patience as I understand what is going on with this really confusing situation. No other resource in Azure has been nearly as convoluted and involved as this. It's unclear why so much effort is being made to prevent users from actually using your resources.

So, if I understand correctly, my Preview environment was approved but given zero quota? I understand having to follow the decision, but that makes sense only if I actually have quota and can use the subscription in some capacity. As it stands now, my subscription has zero quota for the "approved" models and cannot be used at all.

This makes no sense to me. Why approve a model for use when no quota is given? It would seem at the very least, a minimum quota would be provided with an "approved" subscription. Otherwise, the subscription is worthless for any deployments.

I would greatly appreciate any confirmation of my understanding, as well as any corrections. Thank you for your continued assistance.
Sai Kiran Mudavath 90 Reputation points Microsoft External Staff Moderator

2026-10-02T01:16:33.47+00:00

Hi @Mike-E-angelo ,

Yes, your understanding of the current state is correct. Your subscription was previously approved for gpt-image-1.5 access, but the Preview subscription currently shows 0 quota for gpt-image-1.5 for both GlobalStandard and DataZoneStandard.

The important distinction is that model access and quota are separate parts of the deployment process. Microsoft documents gpt-image-1.5 as a limited-access preview model, while quota is managed separately for deployments.

Therefore, the previous onboarding confirmation establishes that the subscription was granted access to the model; it does not, by itself, establish that a non-zero deployment quota is currently allocated. In your case, the current 0 quota is what is causing the InsufficientQuota deployment failure.

Microsoft documents that quota increase requests are evaluated individually and may be denied. Since you have already submitted multiple quota requests and they were denied, there is no documented portal or CLI setting that can manually allocate that missing quota.

So, based on the information currently available, your interpretation is correct: the subscription can have model access while still having no deployment quota available for that model. The reason the quota was not allocated to the Preview subscription despite the earlier onboarding approval is not something the public Q&A documentation allows us to determine.

Thank you!

Answer 1

Mike-E-angelo 631

User's image

I followed the "answer" above and do not see any way of increasing a quota increase. Furthermore the quota was under "Settings" which is not listed in the "answer." Finally, calling az cognitiveservices usage list -l <region> results in a 157KB file and I am not sure how to parse this?

Mike-E-angelo 631 Reputation points

2026-09-30T09:48:45.8533333+00:00

Yes ok adjust filters... nothing there about AI/Foundry:

Answer 2

The error indicates that the subscription has 0 available quota and 0 total quota for Requests Per Minute - GPT Image 1 Mini Generation in the target region. Since az cognitiveservices account list-deleted returns no deleted accounts, there is nothing to purge, so this is not a stale-deleted-resource case.

Use this path:

Confirm quota in the target region Run:
```
   az cognitiveservices usage list -l <region>
```
This shows quota usage for the region.
Request a quota increase for that region In Azure portal:
- Go to Subscriptions
- Select the subscription
- Select Usage + quotas
- Select Request increase
- Submit the increase for the region where the deployment is failing
Quota is regional, so the request must target the same region as the deployment.
If deployment is flexible, change region or model/version If another region has unused quota, deploy there instead. A different model version that fits within current limits can also avoid the failure.
Retry deployment after quota is granted The deployment succeeds only after the quota page shows a nonzero limit for that quota in the target region.

Relevant note on deleted resources: quota remains unavailable for up to 48 hours only when a resource was deleted programmatically and still needs purge. In this case, the deleted-resource list is empty, so the immediate issue is the missing quota allocation itself, not a blocked purge.