Docling taking huge time or crashes sometime to process large and complex pdf

[### Bug
I am trying to use docling with a complex pdf with 180 pages (digital) with lots of images. size 2.7 MB
As this is not scanned images, I have disabled OCR, but I need table extraction and image understanding. So I am using openai VLM for image descriptions.

In this case, docling is taking more than 20 mins to process the document.

This feature is required in my application on frontend, so user experince perspective 20 mins are long time to process the document and user has to wait for 20 mins to ask question based on the uploaded document.

Any suggestions for improvement ?

.
...

### Docling version
2.68.0
...

### Python version
3.13
...

](https://github.com/docling-project/docling/issues)

No source file, test, or entry point is named. Reproduce the report with Docling 2.68.0, Python 3.13, and a comparable 180-page PDF using table extraction and OpenAI VLM image descriptions; identify whether the delay or crash is in PDF processing, table extraction, or image understanding, and define completion as a reproducible performance or stability improvement.

Written by the indexing model from the issue text.