URGENT: Azure OpenAI Services interrupted and mostly not available in Sweden Central

Question

URGENT: Azure OpenAI Services interrupted and mostly not available in Sweden Central

Anonym

Just like yesterday, January 27, at around 09:00 UTC, today, on January 28 at around 09:00 UTC, all communications with GPT-4.1 and GPT-5.x models via the Responses-API (/v1) is interrupted!

So far, Microsoft hasn't recgonized this as a severe service interruption on the Azure Service Health website. Can someone at Microsoft please escalate this?

Anonym

2026-01-28T14:59:38.0733333+00:00

@SRILAKSHMI C thank you. We find it a bit troubling that this issue is still not coming up in any of the Azure Health Service dashboards. So far we can tell that some of our applications that require GPT-4.1 and GPT-5.x in Sweden Central are mostly working again. The overall service health still seems to be impacted (slow) and partly interrupted though.
SRILAKSHMI C 19,735 Zuverlässigkeitspunkte Externe Microsoft-Mitarbeiter Moderator

2026-02-03T09:23:02.1633333+00:00

Hi Daniel Musil,

The issue was caused by an outage at that time, but it has now been resolved. Could you please try again and let me know if you're still experiencing any problems?

Thank you!
Anonym

2026-02-03T09:37:33.8066667+00:00

In the past two days we didn't experience any more service interruptions.
SRILAKSHMI C 19,735 Zuverlässigkeitspunkte Externe Microsoft-Mitarbeiter Moderator

2026-02-04T05:55:45.5933333+00:00

Hi Daniel Musil,

Thanks for confirming the solution. Since I’ve converted my earlier comment into an answer, could you please take a moment to mark it as Accepted? This helps others in the community with the same question find the solution more easily.

Thank you!

Antwort, die vom Frageautor angenommen wurde

SRILAKSHMI C 19,735 Externe Microsoft-Mitarbeiter Moderator

Hello Daniel Musil,

Welcome to Microsoft Q&A and Thank you for reaching out.

As per Team, what happened?

Between 09:22 UTC and 16:12 UTC on 27 January 2026, a platform issue resulted in an impact to the Azure OpenAI Service in Sweden Central region. Impacted customers may have seen HTTP 500/503 errors, failed inference requests, and issues with model deployment metadata. This issue also affected downstream AI Services dependent on Azure OpenAI in this region.

What do we know so far?

Our initial investigation indicates that the issue may be related to elevated error handling within one of our production model dependencies. This temporarily affected request processing, specifically authorization of the incoming requests, causing intermittent service degradation impacting request success rates. We mitigated the issue by stabilizing traffic flow and adjustments to improve request handling and resilience, validating system health, and monitoring recovery to ensure normal operation has been restored.

How did we respond?

09:22 UTC on 27 January 2026 – The issue was detected through service monitoring which is also when customers began to see intermittent availability issues were observed.

12:36 UTC on 27 January 2026 – Initiated mitigation to restart the IRM service on the Sweden Central clusters.

12:46 UTC on 27 January 2026 – Identified that Sweden Central cluster is seeing pods crashing with out-of-memory errors.

13:02 UTC on 27 January 2026 – Initiated mitigation workflow by scaling out nodes in the cluster to improve request handling and resilience.

15:30 UTC on 27 January 2026 – Started to increase the memory available in the pods to alleviate memory load on the cluster.

15:53 UTC on 27 January 2026 – Completed increase in memory in the pods to alleviate memory load on the cluster.

16:12 UTC on 27 January 2026 – Service(s) restored, and customer impact mitigated.

What happens next?

Our team will be completing an internal retrospective to understand the incident in more detail. Once that is completed, generally within 14 days, we will publish a Post Incident Review (PIR) to all impacted customers.

To get notified if a PIR is published, and/or to stay informed about future Azure service issues, make sure that you configure and maintain Azure Service Health alerts – these can trigger emails, SMS, push notifications, webhooks, and more: https://aka.ms/ash-alerts For more information on Post Incident Reviews, refer to https://aka.ms/AzurePIRs The impact times above represent the full incident duration, so are not specific to any individual customer. Actual impact to service availability may vary between customers and resources – for guidance on implementing monitoring to understand granular impact: https://aka.ms/AzPIR/Monitoring

Thank you!

0 Kommentare

1 zusätzliche Antwort

Ihre Antwort

Anonym

2026-01-28T14:59:38.0733333+00:00

@SRILAKSHMI C thank you. We find it a bit troubling that this issue is still not coming up in any of the Azure Health Service dashboards. So far we can tell that some of our applications that require GPT-4.1 and GPT-5.x in Sweden Central are mostly working again. The overall service health still seems to be impacted (slow) and partly interrupted though.
SRILAKSHMI C 19,735 Zuverlässigkeitspunkte Externe Microsoft-Mitarbeiter Moderator

2026-02-03T09:23:02.1633333+00:00

Hi Daniel Musil,

The issue was caused by an outage at that time, but it has now been resolved. Could you please try again and let me know if you're still experiencing any problems?

Thank you!
Anonym

2026-02-03T09:37:33.8066667+00:00

In the past two days we didn't experience any more service interruptions.
SRILAKSHMI C 19,735 Zuverlässigkeitspunkte Externe Microsoft-Mitarbeiter Moderator

2026-02-04T05:55:45.5933333+00:00

Hi Daniel Musil,

Thanks for confirming the solution. Since I’ve converted my earlier comment into an answer, could you please take a moment to mark it as Accepted? This helps others in the community with the same question find the solution more easily.

Thank you!

Answer 1

Anonym

It's January 29. This is the THIRD day in a row on which GPT-4.1 and GPT-5.x models in Sweden Central have service interruptions. None of these issues even appear on the Azure Service Health sites. How is this possible!?

0 Kommentare