Latency in Document Intelligence extremely high

Question

Latency in Document Intelligence extremely high

Stephan Kuhn 30

Hello,

since yesterday morning around 09:00 central european time (CET) the document intelligence services are giving no response anymore. Sending requests to it works fine but the status will stay on running for several hours, now even days. Before it would run around 20 seconds to 2 minutes but not longer.

We are using custom extractions and custom classification models.

When can we expect this to work properly again? As our customers are growing more and more inpatient.

Thanks and kind regards!

Alex Burlachenko 22,120 Reputation points MVP Volunteer Moderator

2025-07-09T09:16:30.5733333+00:00
Stephan hi,

check if its a regional thing. sometimes azure services act up in specific regions. u can peek at the azure status page here. no fancy colors? good. red flags? well... .

for custom models, try reducing the batch size if u're processing multiple docs at once. smaller chunks = less waiting. microsoft docs mention this trick here. worth looking into )

also, give the api a lil' timeout tweak. if it's set too low, it might just... give up. bump it up to something like 300 seconds and see if that helps. this might help in other tools too, not just azure )

as well check this )) are u hitting rate limits? custom models can be greedy with resources. microsoft quietly admits it here. if u're blasting too many requests, the system might just nap on u.

general tip? always monitor those api calls. azure's got built-in tools for that. u can spot if something's choking before customers start yelling. aha ))

and hey... if nothing works, try redeploying the model. sometimes they just need a fresh start, like us after too much wine ) microsoft's docs whisper about it here.

hope this helps )) let us know if the beast is tamed!

Best regards,

Alex

and "yes" if you would follow me at Q&A - personaly thx. P.S. If my answer help to you, please Accept my answer

https://ctrlaltdel.blog/
Stephan Kuhn 30 Reputation points

2025-07-09T09:26:56.99+00:00

Hi,

There are no health issues reported of the service for West Europe where we're at. Other issue tracker sites are also reporting many users having issues with Document Intelligence currently, even though Microsoft doesn't acknowledge it.

Additionally even in the Azure Document Intelligence Studio, when testing the model, its having the same issue.

So no I don't think this issue is on our side but is instead on Microsofts side.
Pavankumar Purilla 11,575 Reputation points Microsoft External Staff Moderator

2025-07-09T09:40:58.58+00:00

Hi Stephan Kuhn,
We understand the impact this issue is having on your operations and sincerely apologize for the inconvenience caused.

As the latency issue appears to be originating from Azure, we would like to raise a support request with Microsoft on your behalf to help expedite the resolution. We have requested the necessary details via private message kindly share the information there at your earliest convenience.

Once we receive the details, we will escalate the issue to the product team for further investigation and keep you informed on the progress.

Thank you for your cooperation and continued patience.

Answer accepted by question author

0 additional answers

Your answer

Alex Burlachenko 22,120 Reputation points MVP Volunteer Moderator

2025-07-09T09:16:30.5733333+00:00

Stephan hi,

check if its a regional thing. sometimes azure services act up in specific regions. u can peek at the azure status page here. no fancy colors? good. red flags? well... .

for custom models, try reducing the batch size if u're processing multiple docs at once. smaller chunks = less waiting. microsoft docs mention this trick here. worth looking into )

also, give the api a lil' timeout tweak. if it's set too low, it might just... give up. bump it up to something like 300 seconds and see if that helps. this might help in other tools too, not just azure )

as well check this )) are u hitting rate limits? custom models can be greedy with resources. microsoft quietly admits it here. if u're blasting too many requests, the system might just nap on u.

general tip? always monitor those api calls. azure's got built-in tools for that. u can spot if something's choking before customers start yelling. aha ))

and hey... if nothing works, try redeploying the model. sometimes they just need a fresh start, like us after too much wine ) microsoft's docs whisper about it here.

hope this helps )) let us know if the beast is tamed!

Best regards,

Alex

and "yes" if you would follow me at Q&A - personaly thx. P.S. If my answer help to you, please Accept my answer

https://ctrlaltdel.blog/
Stephan Kuhn 30 Reputation points

2025-07-09T09:26:56.99+00:00

Hi,

There are no health issues reported of the service for West Europe where we're at. Other issue tracker sites are also reporting many users having issues with Document Intelligence currently, even though Microsoft doesn't acknowledge it.

Additionally even in the Azure Document Intelligence Studio, when testing the model, its having the same issue.

So no I don't think this issue is on our side but is instead on Microsofts side.
Pavankumar Purilla 11,575 Reputation points Microsoft External Staff Moderator

2025-07-09T09:40:58.58+00:00

Hi Stephan Kuhn,
We understand the impact this issue is having on your operations and sincerely apologize for the inconvenience caused.

As the latency issue appears to be originating from Azure, we would like to raise a support request with Microsoft on your behalf to help expedite the resolution. We have requested the necessary details via private message kindly share the information there at your earliest convenience.

Once we receive the details, we will escalate the issue to the product team for further investigation and keep you informed on the progress.

Thank you for your cooperation and continued patience.

Answer 1

Pavankumar Purilla 11,575 Microsoft External Staff Moderator

Hi Stephan Kuhn,
Initially the issue was Latency in Document Intelligence extremely high

We received an update from the internal team that they have fixed the custom pods issue in West Europe for the preview version. This should now be resolved in the new version.

Hope this helps. Do let us know if you have any further queries.

If this answers your query, do click Accept Answer and Yes for was this answer helpful.

0 comments

Share via

Latency in Document Intelligence extremely high

0 additional answers

Your answer