Document Intelligence models running slow/not finishing

Question

Document Intelligence models running slow/not finishing

Bill Corrigan 30

Our Document Intelligence (DI) models are all running very slow, or not completing at all. We push the documents to storage and DI via Power Automate, the flows are all timing out because the models do not finish analyzing. I've also taken one of the documents and tried to perform a test analysis with DI.

The file is small(-ish), 500kb. So file size should not be a problem. None of the projects or models were changed today, so no reason that should have caused the problem. I did enable autoscale this afternoon after the problems started, but it does not seem to have made any improvement to performance.

Azure Status shows no problems, but is that accurate? I cannot find any sort of dashboard that would show me the processes running on our models at the current time. I don't know if too many are running at once, or if we hit a limit for the month. Or, is something else going on with Document Intelligence today?

Prashanth Veeragoni 4,030 Reputation points Microsoft External Staff

2025-04-30T19:55:44.5233333+00:00

Hi Bill Corrigan,

There is an ongoing latency issue for Azure Document Intelligence models, and our internal team is working on this issue. will update you once the issue is resolved in the backend.

Mean time can you please share the region in which you are deploying your model.

Thank you!
Bill Corrigan 30 Reputation points

2025-04-30T20:08:06.8433333+00:00

We're on the East US region.
Manas Mohanty 3,210 Reputation points Microsoft External Staff

2025-05-01T06:12:27.1933333+00:00

Hi Bill Corrigan,

Product team just released a fix for the latency issue; I just tested on my end - East US with 3.1GA and 4.0 API version for prebuilt models with sample and local documents

It works fine to me, please have a try and let us know the model base upon which your custom models were trained if the issue still persists at your side.

I hope this helps.

Thank you.
Bill Corrigan 30 Reputation points

2025-05-01T11:06:16.58+00:00

Everything seems to be moving as expected now. Thank you.
Manas Mohanty 3,210 Reputation points Microsoft External Staff

2025-05-01T12:25:53.43+00:00

Hi Bill Corrigan,

Here is summary of case.

Issue:

Prebuilt and custom models were slow in East Us region and throwing internal server error.

Solution:

Product team just released a fix for the latency issue. It has been conveyed to customer once tested at support side

Status on 1st May 4:36 PM UTC+0530 pm

Model latency is back to normal at customer side.

Thank you.
Philip Traugott 10 Reputation points

2025-05-01T13:50:15.2833333+00:00

We are starting to see failures again, although many are success.
Philip Traugott 10 Reputation points

2025-05-01T13:52:25.1333333+00:00
Prashanth Veeragoni 4,030 Reputation points Microsoft External Staff

2025-05-02T17:46:06.34+00:00

Hi Philip Traugott,

Our product team has confirmed that the backend fix for the latency and failure issues has been fully deployed. Based on our internal testing, everything is now functioning as expected.

Given this, could you please recheck on your end and let us know if you're still encountering any failures? If issues persist, please share a few recent request IDs or logs so we can investigate further.

Thank you!
Bill Corrigan 30 Reputation points

2025-05-02T17:52:16.4866667+00:00

Everything looks fine on my end, but we did not re-experience any of the problems yesterday either. Our flows and models all ran and completed as expected.
Prashanth Veeragoni 4,030 Reputation points Microsoft External Staff

2025-05-02T17:58:25.49+00:00

Hi Bill Corrigan,

Thanks for confirmation, everything works as expected I will repost your comment to answer could you please up-vote and accept it, so that it will be beneficial to others in the community.
Philip Traugott 10 Reputation points

2025-05-02T18:29:36.34+00:00

It has been working for us as well since 10:15am CST yesterday. Thanks!

Accepted answer

0 additional answers

Your answer

Prashanth Veeragoni 4,030 Reputation points Microsoft External Staff

2025-04-30T19:55:44.5233333+00:00

Hi Bill Corrigan,

There is an ongoing latency issue for Azure Document Intelligence models, and our internal team is working on this issue. will update you once the issue is resolved in the backend.

Mean time can you please share the region in which you are deploying your model.

Thank you!
Bill Corrigan 30 Reputation points

2025-04-30T20:08:06.8433333+00:00

We're on the East US region.
Manas Mohanty 3,210 Reputation points Microsoft External Staff

2025-05-01T06:12:27.1933333+00:00

Hi Bill Corrigan,

Product team just released a fix for the latency issue; I just tested on my end - East US with 3.1GA and 4.0 API version for prebuilt models with sample and local documents

It works fine to me, please have a try and let us know the model base upon which your custom models were trained if the issue still persists at your side.

I hope this helps.

Thank you.
Bill Corrigan 30 Reputation points

2025-05-01T11:06:16.58+00:00

Everything seems to be moving as expected now. Thank you.
Manas Mohanty 3,210 Reputation points Microsoft External Staff

2025-05-01T12:25:53.43+00:00

Hi Bill Corrigan,

Here is summary of case.

Issue:

Prebuilt and custom models were slow in East Us region and throwing internal server error.

Solution:

Product team just released a fix for the latency issue. It has been conveyed to customer once tested at support side

Status on 1st May 4:36 PM UTC+0530 pm

Model latency is back to normal at customer side.

Thank you.
Philip Traugott 10 Reputation points

2025-05-01T13:50:15.2833333+00:00

We are starting to see failures again, although many are success.
Philip Traugott 10 Reputation points

2025-05-01T13:52:25.1333333+00:00
Prashanth Veeragoni 4,030 Reputation points Microsoft External Staff

2025-05-02T17:46:06.34+00:00

Hi Philip Traugott,

Our product team has confirmed that the backend fix for the latency and failure issues has been fully deployed. Based on our internal testing, everything is now functioning as expected.

Given this, could you please recheck on your end and let us know if you're still encountering any failures? If issues persist, please share a few recent request IDs or logs so we can investigate further.

Thank you!
Bill Corrigan 30 Reputation points

2025-05-02T17:52:16.4866667+00:00

Everything looks fine on my end, but we did not re-experience any of the problems yesterday either. Our flows and models all ran and completed as expected.
Prashanth Veeragoni 4,030 Reputation points Microsoft External Staff

2025-05-02T17:58:25.49+00:00

Hi Bill Corrigan,

Thanks for confirmation, everything works as expected I will repost your comment to answer could you please up-vote and accept it, so that it will be beneficial to others in the community.
Philip Traugott 10 Reputation points

2025-05-02T18:29:36.34+00:00

It has been working for us as well since 10:15am CST yesterday. Thanks!

Answer 1

Hi Bill Corrigan,

I'm glad that your issue has been resolved and thank you for posting your solution so that others experiencing the same thing can easily reference this! Since the Microsoft Q&A community has a policy that "The question author cannot accept their own answer. They can only accept answers by others ", I'll repost your solution in case you'd like to accept the answer.

Ask: Document Intelligence models running slow/not finishing

Solution: The issue is resolved. That Everything looks fine on your end, but we did not re-experience any of the problems yesterday either. your flows and models all ran and completed as expected.

If I missed anything please let me know and I'd be happy to add it to my answer, or feel free to comment below with any additional information.

If you have any other questions, please let me know. Thank you again for your time and patience throughout this issue.

Please don’t forget to Accept Answer and Yes for "was this answer helpful" wherever the information provided helps you, this can be beneficial to other community members.

Accept answers on Microsoft Q&A | Microsoft Learn

An accepted answer is the answer that the person who asked the question chooses as the one they think best solves their problem.

Thank you!

Share via

Document Intelligence models running slow/not finishing

0 additional answers

Your answer