PdfRecognitionMode.FLOW produces identical output to TEXTBOX (v25.12)

Hey, still playing with the GroupDocs.conversion python package, and found another issue.

We’re seeing PdfRecognitionMode.FLOW have no effect when converting PDF to DOCX using groupdocs-conversion-net v25.12.

Setting FLOW vs TEXTBOX on WordProcessingConvertOptions produces identical output — the resulting DOCX puts all the text in text boxes in both cases,
making the document unusable for editing.

We confirmed the mode does reach the .NET layer (FLOW takes ~3× longer to run), but the word/document.xml inside the output DOCX is byte-for-byte identical between the two modes.

We reproduced this both via the Python wrapper and by calling GroupDocs.Conversion.dll directly from C#, ruling out a wrapper issue.

Notably, converting the same PDF through the GroupDocs online app produces correct output with no text boxes.

Is this a known issue in v25.12, and is there a workaround or fix available?