Only first 2 pages analyzed in document intelligence

Hello Flavius Sfirlea,

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

This behavior is actually a very common point of confusion with Document Intelligence, and the good news is that your S0 upgrade is not the issue.

Even on paid tiers like S0, the Document Intelligence Studio (web UI) only displays OCR results for the first two pages when you are labeling or testing a custom model inside the Studio. This is a UI limitation of the labeling/preview tool not a processing or billing limitation.

Why you still see only 2 pages in Studio

The Studio intentionally shows only the first two pages to keep the interface responsive and avoid performance issues with large documents. This has no impact on what the model actually processes.

What happens behind the scenes

Even though the UI displays only two pages:

• Your custom model trains on all pages in your training documents.
• Your model analyzes all pages during actual inference.
• Full-page processing works normally when using the REST API, SDKs, or Postman.

What you can check

To ensure everything is working end-to-end:

• Try running the same file using the API/SDK you should see the complete multi-page output.
• Make sure you’re using a Custom Extractor or Custom Neural model, as prebuilt models have different limits.
• Confirm that the Page Range is not set to “1–2” during testing.
• Test with a fresh document in case the Studio is showing cached results.

If the API response itself only returns two pages, then it would indicate a separate problem. But based on your description, this looks to be only the Studio preview limitation.

If you share which model type you're using and whether you tested via API, I can help you verify the full output.

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

Thank you!