Changelog for Java
RSS2.1.14 - 15 Nov 2026
General
- AddedAdds data extraction for charts drawn as vectors in PDFs: series, categories, and values, without an AI model. (#3872)
- AddedAdds data extraction for pie and donut charts drawn as vectors in PDFs, without an AI model. (#3872)
- AddedAdds MRZ zones to JSON, Markdown, and IR Lite output, with their format and whether their check digits are valid. (L#XT-617)
- ChangedImproves AI chart extraction by checking its values against the chart’s vector drawing in PDFs. (#3872)
- ChangedImproves chart extraction from born-digital PDFs by reading labels in chart images and outlined text with OCR. (#3872)
- ChangedImproves chart extraction in PDFs whose chart labels are drawn as glyph outlines by reading them with OCR. (#3872)
- ChangedChanges the barcode element type in the JSON output to
machineReadable; it now also covers MRZ zones. (L#XT-617) - ChangedUpdates the cryptography dependency used by the Java SDK’s native library to address denial-of-service vulnerabilities CVE-2026-70162 and CVE-2026-84008. (#3806)
- ChangedChanges chart classification in JSON and IR Lite to the AI, reader, or model chart kind with its probability and source. (#3853)
- ChangedChanges the page marker in Markdown and text output from
nutrient-page-starttodocument-page. (L#XT-840) - FixedFixes a missing or incompatible native library being reported as a generic error. It’s now reported as
NativeLibraryLoadException, orNativeLibraryLoadFailedfor status-code callers, and names the library. (#3781) - FixedFixes garbled Unicode attachment file names when reading MSG emails. (L#MAMBO-2066)
- FixedFixes loading the SDK on Red Hat Enterprise Linux 9, Rocky Linux 9, and AlmaLinux 9. (#4135)
- FixedFixes values of line series drawn against a secondary axis in charts of born-digital PDFs. (#3872)
Imaging
- ChangedOptimizes AI document processing by loading vision models from disk only when an inference session is first created. (#3258)
- ChangedUpgrades bundled image codecs and network libraries to versions that fix published security vulnerabilities. (#3938)
- FixedFixes picture classification on Linux x64, where every picture was classified as other and charts weren’t recognized. (#3854)
OCR
- ChangedImproves handwriting extraction by recovering detected words the AI model didn’t transcribe. (#3370)
- ChangedImproves heading detection and heading levels on scanned pages, which now get the same layout analysis as PDFs with text. (#4024)
- ChangedImproves the reading order, caption detection, and form field grouping of parsed pages. (#3845)
- FixedFixes image-processing warnings written to the standard error output during OCR. (L#OCR-1115)
- FixedFixes local vision model failures on high-resolution scans by scaling oversized image regions to the model context. (#3386)
Office
- FixedFixes chart titles missing from Excel, PowerPoint, and Word documents converted to PDF when the file leaves out the
autoTitleDeletedsetting. Files written by XlsxWriter, Google Sheets, and older Excel versions leave it out. (L#MAMBO-2140) - FixedFixes garbled Hebrew and other non-Latin subjects in older MSG emails that declare their code page. (L#DV-798)
- FixedFixes MSG email bodies being cut off after the first line when preformatted text was displayed. (L#GDPIC-1395)
- FixedFixes legacy VML shapes, picture color modes, transparent picture colors, and grouped shapes looking different from Word and PowerPoint. This affects Office documents converted to PDF. (L#MAMBO-2139)
- AddedAdds a detected or forced FIPS mode through
CryptoSettingsfor supported PDF cryptography. Restrictions on RC4, MD5, and SHA-1 depend on the operation and host module. Windows CNG can compute MD5 and verify existing SHA-1 signatures, but can’t sign with SHA-1. Supported operations use a FIPS 140-3 validated OS module where available. Hardware keys may use another provider, macOS validation depends on its OS module, and PKCS#12 MAC key derivation remains outside the module. (L#NAT-666, L#NAT-681) - AddedAdds PDF signature validation against trusted certificates, with revocation and timestamp checks (ETSI EN 319 102-1). (L#GDPIC-1671)
- AddedAdds flattening of PDF form fields, for a single field, one page, or the whole document. (L#NAT-721)
- AddedAdds HTML conversion controls and private-network allow lists while preserving linked local resources. (#4055)
- ChangedImproves PDF/UA auto-tagging: Captions are tagged as
Caption, and footnotes asNote, referenced from their calls. (L#DWS-1221) - ChangedChanges PDF signing and encryption to the bundled Botan library;
CryptoSettingsselects the OS library for FIPS mode. (#3677) - ChangedChanges PDF redaction to write black-and-white images losslessly, so content outside a redaction area keeps its shape. (#4000)
- FixedFixes PKCS#12 files that demand excessive key-derivation work tying up a signing thread; they’re now refused. (#3677)
- FixedFixes valid digital signatures reported as invalid when their signature data ends with a zero byte. (#3677)
- FixedFixes invalid-content reporting for PDFs whose readable catalog lacks a usable page-tree dictionary. (L#DWS-1272)
- FixedFixes malformed PDF trailers being reported as access-denied errors. (#3901)
- FixedFixes new signature fields always named
Signature1: They take the signer name, with periods as_and_2added if taken. (L#GDPDF-1545) - FixedFixes PDF redaction leaving pixels of inline, undecodable, 4-bit, and rotated images under a redaction area. (#4000)
- FixedFixes PDF redaction removing text that a page draws after a form XObject. (#4000)
- FixedFixes PDF text extraction joining a label with the text it covers and dropping characters from widely spaced rows. (#3847)
- FixedFixes PDF text extraction joining words when a font leaves its blank space glyph out of the Unicode map. (L#XT-827)
- FixedFixes PDF/UA auto-tagging giving images the actual text “Image,” which screen readers read instead of the alt text. (L#DWS-1215)
- FixedFixes PDF/UA auto-tagging joining headings to nearby text or lists, flattening heading levels, and missing ordered lists. (L#DWS-1217, L#DWS-1218, L#DWS-1219)
- FixedFixes PDF/UA auto-tagging tagging running headers, footers, and page numbers as paragraphs instead of artifacts. (L#DWS-1216)
- FixedFixes PDF/UA auto-tagging titling a multipage document after its last page’s heading instead of its first. (L#DWS-1220)
- FixedFixes signature fields below a parent form field being left out when PDF signatures are counted, read, or validated. (L#GDPIC-1671)
- FixedFixes the media-load timeout being ignored when converting EML and MSG emails to PDF. (L#DV-536)
- FixedFixes document text written to HTML and Markdown exports as unescaped, potentially unsafe HTML markup; it’s now escaped by default. (L#GDPIC-1676)
- FixedFixes RSA-PSS signatures with a non-default salt length and DSA signatures over SHA-2 failing verification. (#3677)
Templating
- AddedAdds DOCX subtemplates to Word templates: A placeholder can insert another Word document, filled from the same data. (L#OFTAPI-30)
- FixedFixes Word template processing failing when an HTML value in a header or footer adds the document’s first list. (L#OFTAPI-30)
Vision
- AddedAdds a Markdown
EnableTableCellInlineFormattingsetting that keeps bold, strikethrough, and other inline styles in table cells. (L#XT-635) - AddedAdds a reasoning effort setting to every AI model connection, used by parsing, extraction, and form labeling. (L#XT-653)
- AddedAdds AI model settings to each AI operation, so parsing and extraction can each use their own model. (L#XT-653)
- AddedAdds an
EmitPageBoundaryMarkersoption that marks the start of each page in Markdown output fromparse. (L#XT-840) - AddedAdds an opt-in
FoldTypographicPunctuationsetting that writes dashes and curly quotes as ASCII in Markdown, JSON, and plain-text output. (#3905) - AddedAdds
hyphenAtLineEndto JSON output words, reporting whether a line-end hyphen is a wrap, compound, or suspended hyphen. (#3881) - AddedAdds table row roles: Titles and section labels are written as
<th>in Markdown, and JSON cells get arowRole. (L#XT-611) - AddedAdds the DirectX shader compiler download for GPU inference on Windows on Arm, as on Windows x64. (#3843)
- AddedAdds the text read inside pictures, such as ID cards and pasted receipts, to the JSON, Markdown, and data extraction output. (#4063)
- ChangedImproves chart extraction from PDFs with outlined labels, shared legends, multipanel figures, or printed values. (#3876)
- ChangedImproves OCR accuracy on hard-to-read digits and amounts by rereading the word and checking table totals. (#4136)
- ChangedImproves AI extraction of long tables and bank statements. Each list-pass call now covers at most 120 table or ledger rows. Amounts the text layer lists after their rows are rejoined to them. A list window that times out no longer drops the rows the other windows returned. (#3856)
- ChangedImproves data extraction on documents with struck-through, bold, or superscript text by showing the model the text styles. (L#XT-635)
- ChangedImproves Markdown inline formatting: Strikethrough, underline, bold, and italic can now cover part of a word. (L#XT-827)
- ChangedImproves Markdown output by keeping the line breaks that end a title line, label, list item, or address line inside a paragraph. (#3882)
- ChangedChanges local AI models to follow
VisionAccelerationSettings.Provider = CpuandGpuExcludedModules; they then run on the CPU, which can be much slower. (#3840) - ChangedChanges data extraction and AI classification to use the AI model selected in the vision settings. (L#XT-653)
- ChangedChanges data extraction to mark the citation of an empty value as approximate, since nothing on the page can confirm it. (L#XT-652)
- ChangedChanges Markdown export with inline formatting to keep strikethrough in table cells. (#4003)
- ChangedChanges Vision AI extraction and classification to default to temperature 0 on models that accept it so runs repeat. (L#XT-650)
- ChangedChanges Vision document classification and splitting to fail when
OpenSettings.PageRangeselects no page. Classification with the decision model now also fails when none of the selected pages can be rendered. (#4120) - ChangedEnables local vision-language models to run on the GPU on Windows, and on Linux without an NVIDIA GPU, through Direct3D 12 and Vulkan. They previously ran on the CPU there. (#3830)
- ChangedMoves the parsing prompt, zone strategy, page reference, and batch size from provider settings to parse settings. (L#XT-653)
- RemovedRemoves the AI augmenter settings; the zone classification confidence moves to the parse settings. (L#XT-653)
- RemovedRemoves the environment variables for the AI network timeout and structured-output mode; the timeout is a setting. (L#XT-653)
- FixedFixes the page size and resolution reported for Excel sheets read structurally, which now match the pages the sheet prints on. (#3837)
- FixedFixes AI calls to OpenAI, Anthropic, or Google without an API key being sent instead of failing with the setting to fill in. (L#XT-653)
- FixedFixes AI data extraction with Claude stopping at 4,096 output tokens instead of the configured limit. (L#XT-653)
- FixedFixes AI extraction schema validation not rejecting invalid input before documents were processed. (L#DWS-855)
- FixedFixes chart values read half a label off when the value labels stand on their gridlines instead of being centered on them. (L#XT-620)
- FixedFixes data extraction citations on a block without the value: They now move to the value or are marked approximate. (L#XT-651)
- FixedFixes data extraction failing with confidence scores enabled when a local model samples at a nonzero temperature. (L#XT-649)
- FixedFixes element bounding boxes in parse output so they fit their text, checkboxes, table frames, and pictures. (#4099)
- FixedFixes handwriting correction failing silently with Google and with the built-in AI model. (L#XT-653)
- FixedFixes high-effort OpenAI requests timing out at 300 seconds; their default per-attempt timeout is now 600 seconds. (#4056)
- FixedFixes ID cards and other pictures that came out as a false barcode result instead of a picture. (#4029)
- FixedFixes local AI models crashing on x64 CPUs without AVX-512, and picks the fastest instruction set the CPU supports. (#3957)
- FixedFixes lost text in digital PDFs where a table or chart is an image, including one with a caption or title on it. (L#XT-715)
- FixedFixes Markdown footnotes that lost italic, underline, and superscript formatting, including the footnote numbers. (L#XT-610)
- FixedFixes Markdown inline formatting that left brackets, quotes, and parentheses of bold or italic words outside the markers. (L#XT-606)
- FixedFixes Markdown inline styles being dropped on VLM-corrected text when inline formatting is enabled. (L#XT-859)
- FixedFixes Markdown output that replaced en and em dashes, minus signs, and curly quotes with plain ASCII characters. (L#XT-608)
- FixedFixes Markdown output that wrote a lone bold line, such as a timed program title, as a one-item list. (L#XT-609)
- FixedFixes Markdown strikethrough on redlined PDFs: Deletions printed next to an insertion or inside a word are now marked. (#4003)
- FixedFixes minus and plus signs lost or misread in scanned tables, and improves reading of characters that stand alone. (L#XT-700)
- FixedFixes ONNX-based Vision models staying on the CPU for the rest of the process after a GPU runtime download failed. They stayed on the CPU even once a later retry succeeded. (#3974)
- FixedFixes page orientation detection running on the CPU instead of the NVIDIA GPU. (#3975)
- FixedFixes parse output returning a document title at the top of a page as a page header instead of a title. (L#XT-841)
- FixedFixes QR codes and barcodes missing from Vision results when the picture around them is read as text. (L#XT-756)
- FixedFixes self-hosted AI models thinking by default, which made parsing and data extraction many times slower. (L#XT-653)
- FixedFixes small Data Matrix barcodes in parsed documents being returned as pictures instead of barcodes. (L#DWS-1168)
- FixedFixes table cell confidence always reading 1.0 on scanned pages; it now shows how sure the table model was of each cell. (L#XT-703)
- FixedFixes tables written twice, as text and again at the end of the page, when parsing an image in
structuremode. (#4141) - FixedFixes the fallback AI model receiving the primary model’s temperature, token limit, and reasoning effort. (L#XT-653)
- FixedFixes Vision document classification and splitting processing pages outside
OpenSettings.PageRange. (#4120) - FixedFixes Vision form and language detection processing pages outside
OpenSettings.PageRange. (#3748) - FixedFixes Vision models being reloaded, or failing, when concurrent requests use different
VisionAccelerationSettings.Providervalues. (#3852) - FixedFixes Vision steps using the GPU when the request’s
VisionAccelerationSettings.Provideris set toCpu. This affects steps that take no document settings of their own (orientation correction, handwriting recognition, zero-shot classification, and document splitting). (#3831) - FixedFixes watermark and stamp text that merged into the text blocks and tables of PDF pages. (L#XT-706)
2.1.13 - 22 Oct 2026
General
- AddedAdds in-process local model support for structured extraction without an external model server. (L#NAVI-86)
- AddedAdds a no-sampling decision-readout primitive (
LLAMAW_Decide/LLAMAW_TokenizeSingle) to the in-process llama.cpp wrapper. (#3645) - AddedAdds an opt-in (
enableComplexFieldGrounding) path that scores object/array-row fields through the document decision model’s real JSON context instead of a synthesized heuristic sentence; accuracy impact vs. the existing heuristic isn’t yet independently measured. (#3647) - AddedAdds opt-in local decision-model backends for document classification and splitting. (#3644, #3648, #3649)
- AddedAdds continuation links for paragraphs and tables split across pages in the JSON export and merged Markdown output. (J#NAVI-74)
- AddedAdds
disposeto editors, which closes the document being edited and releases its native resources. (#3422) - AddedAdds highlighted-word metadata and preserves highlights in Markdown and JSON exports. (#3315)
- AddedAdds llama.cpp model discovery for configurable local vision models. (L#NAVI-73)
- AddedAdds optional partial redaction that preserves the final logical group of detected sensitive values. (#3597)
- AddedAdds support for the gpt-6-luna OpenAI vision model. (#3487)
- AddedAdds
Vision.Decide, a generic decision-model API that lets callers submit their own state and questions JSON directly to the native decision readout. (#3646) - AddedAdds text from small pictures and extracted chart data to Vision JSON output. (#3521)
- ChangedImproves document processing performance with SIMD-accelerated image preprocessing and model decoding. (#3327)
- ChangedImproves AI extraction reliability by retrying an OpenAI request once when it times out, instead of failing. (#3580)
- ChangedImproves chart value accuracy by keeping axis titles and legends inside the region sent to the vision model. (#3322)
- ChangedImproves document parsing: Detects code blocks, recognizes numbered section headings, and removes duplicated OCR text. (#3686)
- ChangedImproves Markdown export: plain captions, no generated chart summaries, and valid image syntax for pictures. (#3637)
- ChangedImproves SVG rendering security. (L#NAT-691)
- ChangedImproves table recognition speed when tables are processed on the CPU. (#3582)
- ChangedImproves table extraction by preserving cell text on rotated pages, recovering missing rows in long tables, and detecting tables missed in the initial page analysis. (#3488)
- ChangedChanges the default OpenAI vision model to gpt-6-luna. (#3707)
- ChangedChanges the default per-attempt OpenAI network timeout from 600 to 300 seconds, so a stalled AI call is retried sooner. (#3777)
- RemovedRemoves the embedded OpenSSL; local VLM model downloads now use the operating system TLS stack and proxy settings. (#3530)
- FixedFixes an issue where prompts were truncated in the VLM generation path. (#3645)
- FixedFixes a macOS crash at process exit after running the local in-process vision language model. (L#NAVI-85)
- FixedFixes process crashes when several threads register a license key or evaluate a licensed feature at the same time. (L#GDPIC-1655)
- FixedFixes an issue where trial-mode feature traces were lost and trial initialization ran more than once when licensed features were first used concurrently. (L#GDPIC-1655)
- FixedFixes an issue where the licensing banner and licensed feature list were printed to the console after a production license key was registered. (L#GDPIC-1622)
- FixedFixes an issue where AI data extraction failed with a misleading error when a schema field had an empty list of allowed values. (L#DWS-806)
- FixedFixes an issue where AI extraction failed on very large schemas. Properties are now extracted in groups. (#3580)
- FixedFixes an issue where AI parsing failed a page when the provider’s content filter stopped the VLM reply. Affected zones retain their OCR text. (#3580)
- FixedFixes AI reader requests failing with Claude Opus 5 and 5.5 and OpenAI gpt-6-sol, gpt-6-astra, and gpt-6.1-sol. (L#NAVI-169)
- FixedFixes crashes during concurrent document processing with GPU acceleration on macOS. (#3769)
- FixedFixes CUDA acceleration on Linux systems with glibc older than 2.38, such as Debian 12 and Ubuntu 22.04. (L#NAVI-165)
- FixedFixes document processing failing on pages that need a targeted re-OCR when the OCR engine is set to PaddleOCR. (#3579)
- FixedFixes an issue where editors closed their document twice and leaked native handles when released. (#3422)
- FixedFixes malformed Markdown when extracted text contains less-than signs before HTML tables. (#3473)
- FixedFixes page rendering in document processing to apply the page’s rotation, as the
DocProcessorrender command does. (#3693) - FixedFixes PDF 2.0 saves to preserve existing XMP and migrate standard Info fields when XMP is absent. (#3091)
- FixedFixes signing encrypted PDFs without losing protection and opening certificate-encrypted PDFs. (#3091)
- FixedFixes the macOS native library requiring macOS 26; the minimum supported macOS version is now 14.0. (#3813)
- FixedFixes VLM-based extraction failing or misreading text on high-resolution scans and with newer models. (#3382)
Barcode
- AddedAdds USPS Intelligent Mail Barcode (IMb) recognition to
GdPictureImagingwith theBarcodeIntelligentMailReadermethods. The best quality mode reads barcodes at any angle. AddsGdPictureStatus.BarcodeWrongState, returned when the native barcode reader fails internally. (L#GDBC-498) - ChangedImproves QR code recognition on scans with smudged timing patterns or a squeezed band of rows. (#3542)
- ChangedImproves recognition of QR codes on scans that squeeze rows hard, damage a corner square, or crop the code tightly. (#3545)
Imaging
- AddedAdds TWAIN methods to read the extended image information (
DAT_EXTIMAGEINFO) a scanner reports for each acquired image, including hardware-detected patch codes. (J#GDPIC-1614) - ChangedOptimizes the raster operators used by
DrawGdPictureImageOPandDrawFilledRectangleOPwith SIMD. (#3327) - FixedFixes 32-bit processes reporting a production license as trial and printing the evaluation banner. (J#GDPIC-1663)
- FixedFixes loading of the Linux native libraries on glibc 2.34 systems such as Amazon Linux 2023. (#3514)
- FixedFixes a memory leak occurring on every scanned page when TWAIN barcode detection is enabled. (J#GDPIC-1614)
- FixedFixes form field detection crashing on CPU-only machines when the exported model uses float16 precision. (#3650)
- FixedFixes loading of standalone JBIG2 (
.jb2) image files, including multipage files. (L#GDPIC-1656) - FixedFixes an issue where MP4, MOV, and other video files were detected as HEIF images and failed to open with a generic error. (#3091)
OCR
- AddedAdds automatic handwriting layout recovery that reconstructs under-segmented handwritten regions before extraction. (#3374)
- AddedAdds automatic per-page routing (
VisionEngine.Auto) that picks the cheapest suitable extraction pipeline per page. (#3296) - AddedAdds GPU-accelerated Paddle OCR with automatic fallback and hybrid Tesseract routing for vision document processing. (#3317)
- ChangedImproves automatic OCR language detection. (L#NAVI-115)
- ChangedImproves automatic OCR language selection. (L#NAVI-123)
- ChangedImproves OCR accuracy on GPU machines. (L#NAVI-126)
- ChangedImproves automatic OCR language detection speed. (#3566)
- ChangedImproves bold detection on scanned pages by comparing each word’s stroke weight with same-size text on the page. (#3641)
- ChangedImproves document splitting accuracy on merged files that contain rotated scanned pages. (#3497)
- ChangedImproves OCR language detection on scanned pages, identifying more languages correctly in less time. (#3710)
- ChangedImproves OCR on low-contrast scans. (L#NAVI-108, L#NAVI-109)
- ChangedImproves OCR processing speed when using GPU acceleration. (#3299)
- ChangedImproves OCR speed on GPU by checking uncertain words in parallel. (#3566)
- ChangedImproves page orientation detection on scanned pages. (#3710)
- ChangedImproves text extraction from scanned pages that carry a few typed or stamped words by reading the whole page with OCR. (#3710)
- ChangedKeeps the text of table regions that the recognized cell structure doesn’t account for: Words inside a detected table but outside every recognized cell (total lines under the last row, lines merged into a neighboring row) are now exported as text after the table instead of being dropped. (#3342)
- ChangedMakes document conversion reproducible: Reading order, spatial lookups, and word ordering no longer depend on the order in which internal collections happen to enumerate, so converting the same document twice produces the same text. (#3343)
- ChangedChanges document data extraction to ignore a PDF’s invisible OCR text layer and read those pages with its own OCR. (#3628)
- ChangedEnables fully local structured extraction: Extract with provider
localand no endpoint now runs the model in-process, downloading the default local model on first use, instead of requiring an OpenAI-compatible server. (L#NAVI-166) - ChangedOptimizes OCR processing speed on scanned pages. (#3710)
- ChangedOptimizes text detection speed in OCR and language detection, most on GPU. (J#NAVI-218)
- ChangedUpdates the in-process local model runtime to llama.cpp 0.5.0, which adds support for newer local models such as Qwen3.8 (models with multitoken prediction layers). (L#NAVI-167)
- FixedFixes recognized text being dropped from the output when it falls in a region no recognition stage produced content for. (#3343)
- FixedFixes an issue where label and value words on the same row of a scanned page could be emitted out of order. (L#NAVI-84, #3343)
- FixedFixes a crash when OCR with several languages reads a word differently in each language. (#3786)
- FixedFixes automatic OCR language detection misreading Latin-script pages as Cyrillic when
Ocr.maxScriptsis above 1. (L#NAVI-148) - FixedFixes making PDFs searchable in Danish, Swedish, Norwegian, and most other languages, which were read as English. (L#NAPY-24)
- FixedFixes multizone OCR returning no words for a zone that shares a text line with an earlier zone. (#3609)
- FixedFixes OCR failing on every page when the OCR engine is pinned to Paddle. (#3423)
- FixedFixes thousands separators in amounts read from faint scans. (L#NAVI-135)
Office
- AddedAdds support for the
STYLEREFfield when converting Word documents to PDF, so a header or footer prints the text of the relevant paragraph — typically the current heading — on each page instead of repeating the value cached in the document. (J#MAMBO-2116) - AddedAdds Hebrew and Arabic letter list numbering to DOCX-to-PDF conversion. (J#MAMBO-2129)
- ChangedImproves the precision of cell and paragraph shading when converting documents to PDF and SVG. (J#MAMBO-2126)
- FixedFixes HTML and Markdown export of Office documents and images, which failed instead of converting them to PDF first. (#3091)
- FixedFixes missing rows and incorrect row heights when converting presentation tables to PDF. (J#MAMBO-2126)
- FixedFixes Excel workbooks failing to open when a table style omits its optional display settings. (#3528)
- FixedFixes Excel workbooks with damaged internal references failing to open when they could be repaired. (#3528)
- FixedFixes MSG files rendering as binary garbage when their compound file directory is fragmented. (J#DV-759)
- FixedFixes a paragraph after a floating table being pushed below the table by unneeded automatic spacing. (L#MAMBO-2131)
- FixedFixes a table of contents entry or other field result rendering entirely in the font of its first character. (L#MAMBO-1911)
- FixedFixes field results and field codes being swapped when converting legacy Word documents to Word format. (L#MAMBO-1911)
- FixedFixes Markdown-to-PDF conversion producing raster pages with no extractable text. (L#GDPIC-1660)
- FixedFixes multilevel list numbers being scrambled (6.1 shown as 16.) in right-to-left DOCX-to-PDF conversion. (J#MAMBO-2129)
- FixedFixes pictures placed in a line of text being pinned to a corner when converting legacy Word documents. (L#MAMBO-1911)
- FixedFixes PowerPoint tables using the “No Style, No Grid” table style being rendered with black cell borders. (L#MAMBO-2132)
- FixedFixes regular text rendered in a heavy weight when converting Office documents that use static multiweight font families, such as Lato. (#3699)
- FixedFixes right-to-left text order and column-break layout when converting Word documents to PDF. (J#MAMBO-2045)
- FixedFixes right-to-left text running under floating text boxes and images instead of wrapping around them in DOCX to PDF. (L#MAMBO-2130)
- FixedFixes shapes with a picture or texture fill disappearing when converting PowerPoint presentations to PDF. (J#MAMBO-2121)
- FixedFixes table columns in Word documents being too narrow for their longest word, so words no longer break mid-word. (L#MAMBO-2131)
- FixedFixes Word documents failing to open with
InvalidParameterwhen a table cell spans zero grid columns. (L#MAMBO-2136) - FixedFixes Word table cells with zero cell margins being rendered with the table’s default margins. (L#MAMBO-2131)
- FixedFixes words being repeated or truncated in a table of contents rendered with small capitals. (L#MAMBO-1911)
- ChangedImproves PDF encryption to use random AES initialization vectors and keys and the system cryptographic provider. (#3665)
- ChangedImproves heading detection for document titles in Markdown output. (#3567)
- ChangedImproves table detection in PDF layout analysis. (#3567)
- ChangedImproves detection of ruled tables with stacked records, merged cells, or shaded rows in PDF layout analysis. (#3520)
- ChangedImproves font handling for specific document configurations. (J#PDFA-114)
- ChangedImproves PDF-to-HTML conversion speed and memory use for documents with many text runs. (#3631)
- ChangedImproves text extraction fidelity for ligature glyphs, underlines, and clipped text on rotated PDF pages. (#3488)
- ChangedImproves text extraction for fonts that identify characters by glyph name. (#3653)
- ChangedImproves the accuracy of heuristic document layout detection. (#3377)
- FixedFixes HTML and PDF export replacing the underlying error with a generic one, hiding the real cause of a failed export. (#3091)
- FixedFixes annotation text such as contents and author reading back garbled when stored as UTF-16 or non-ASCII. (#3505)
- FixedFixes bold text extracted as regular from PDFs whose embedded fonts carry no weight information other than the font header. (#3629)
- FixedFixes page rendering failing on PDFs that embed CFF2 OpenType fonts; affected text now renders with a substitute font. (#3579)
- FixedFixes PDF 2.0 text stored as UTF-8 — such as annotation contents, form field values, and bookmarks — reading back garbled. (#3505)
- FixedFixes PDF text extraction for Type 1 fonts with a scaled font matrix and for Euro signs mapped to a wrong character. (#3686)
- FixedFixes PDF text extraction to use the actual text a PDF gives for glyphs and to return ligatures such as fl as letters. (J#GDPDF-1537)
- FixedFixes redaction region placement on PDF pages whose page box doesn’t start at the origin. (L#GDPIC-1658, L#GDPDF-1532)
- FixedFixes rotated text, such as vertical table headers and axis labels, missing from PDF layout extraction. (#3625)
- FixedFixes structured data extraction failing for schemas that use local
$defsor$refdefinitions. (#3580) - FixedFixes symbolic fonts that name glyphs by Unicode extracting raw character codes instead of the characters. (#3625)
- FixedFixes text drawn with a negative font size in a half-turned text matrix being extracted mirrored and rendered with misplaced spacing. (J#GDPDF-1344)
- FixedFixes the PDF reducer clearing the producer metadata of the output document when no producer name is set. (#3436)
- FixedFixes title metadata preservation and annotation placement when editing PDF tags. (#3627)
- FixedFixes XFDF export failing on FIPS-enabled hosts when the PDF contains file attachment annotations. (#3535)
Templating
- FixedFixes line breaks and tabs being dropped from values substituted into Office document templates. (J#OFTAPI-27)
- FixedFixes Office template processing failing when a section wraps a paragraph loop inside a single table cell. (J#OFTAPI-25)
Vision
- AddedAdds a dedicated connection setting for the form field detect-extract model, plus an optional fallback flow and connection for form field enrichment. (#3452)
- AddedAdds a Markdown
EmitFigurePlaceholderssetting; the bare[Figure]picture placeholder is now off by default. (L#NAT-626) - AddedAdds a
vlm_batches_failedevent to VLM-enhanced recognition that reports how many vision-model batches failed on a page and the first error, so a page that lost part of its text to a provider failure is visible in the logs instead of passing silently. (#3448) - AddedAdds an opt-in
enableInlineFormattingoption toDWS.Parsefor bold, italic, and other inline styles in Markdown. (L#NAT-514) - AddedAdds an opt-in setting that reads a list printed in side-by-side column groups with a repeated header as one table. (L#NAVI-187)
- AddedAdds description, value, and group ID to each detected form field in the JSON output, and reports the enrichment strategy used to produce each form’s fields. (#3452)
- AddedAdds in-process document parsing and schema-based data extraction through
DataExtraction, with options configured through document settings. (#3424, #3608) - AddedAdds orientation correction to the Structure engine for rotated pages that require OCR. (#3555)
- AddedAdds
streamCandidateLabelsto document splitting: per-page and per-segment document types, with a split at each type change. (#3649) - AddedAdds support for reading USPS Intelligent Mail barcodes in parsed documents, such as the barcode printed on an envelope or mail label. (L#DWS-869)
- AddedAdds the font class to text from born-digital PDFs and cleans up their font family names in Vision output. (#3751)
- AddedAdds the font family, class, and size of each paragraph and table cell to the Vision JSON output. (#3717)
- ChangedImproves Markdown export by rejoining words hyphenated across line breaks using language-specific dictionaries. Disable
joinHyphenatedLineBreaksto preserve those breaks. (L#NAT-627) - ChangedImproves Excel-to-Markdown conversion by recognizing each table, heading, and paragraph on a sheet. (#3528)
- ChangedImproves checkbox and option-state extraction with VLM-enhanced recognition: Checkboxes, radio buttons, and tick boxes are transcribed in place as marked or empty boxes and are no longer lost when OCR and VLM text are merged. (#3447)
- ChangedImproves OCR accuracy on scanned pages. (L#NAVI-130)
- ChangedImproves OCR accuracy when the recognition pipeline runs directly: It now reads scans as the extraction modes do. (#3607)
- ChangedImproves the reading order of scanned forms and receipts: A page the layout detector splits into many short one-line regions is now read row by row, which keeps each label next to its value. (#3450)
- ChangedImproves VLM-enhanced recognition accuracy. (L#NAVI-124)
- ChangedImproves word accuracy in VLM-enhanced recognition. (L#NAVI-125)
- ChangedImproves
AdaptiveOCRtable output: rejects layout boxes misread as tables and emits enumerated items as lists. (#3520) - ChangedImproves AI data extraction from charts and diagrams with OpenAI models: The images of the pages that hold a chart or a picture are now sent to the model with their text by default. Set
includePageImagesto send every page. (#3782) - ChangedImproves AI data extraction of long lists on large documents by filling them from a second pass over smaller page windows. (#3778)
- ChangedImproves AI extraction from very long or wide pages, such as well logs and time-location charts, with gpt-6 on OpenAI: Such a strip page is now sent as overlapping image tiles near its rendered resolution instead of one image the provider shrinks. (#3803)
- ChangedImproves automatic document parsing speed by using available GPUs for page routing. (L#DWS-967)
- ChangedImproves bold and italic detection on scanned pages, and adds font family, class, and size to recognized text. (#3717)
- ChangedImproves chart data extraction with fast geometry-based bar chart reconstruction. (#3316)
- ChangedImproves computed values in AI data extraction, such as totals, counts, and shares: A second pass works out the number fields the model left at 0, and OpenAI reasoning models now default to a high reasoning effort. (#3783)
- ChangedImproves form field detection and fill-state accuracy with an updated detector and a dedicated fill-state model. (#3330)
- ChangedImproves ICR table extraction for tables without printed grid lines, cut across page tiles, or read as text or form. (#3625)
- ChangedImproves ICR table headers, columns, and splits, and keeps grouped header rows in Markdown table export. (#3625)
- ChangedImproves table detection in VLM-enhanced ICR by asking the VLM whether grid-like text regions are tables. (#3625)
- ChangedImproves table extraction on scanned documents. (L#NAVI-106, L#NAVI-107)
- ChangedImproves table text in VLM-enhanced recognition. (L#NAVI-110)
- ChangedImproves text recognition in tables, including tables pasted as images and text on shaded cells. (#3610)
- ChangedImproves the reliability of model downloads. (L#NAVI-117)
- ChangedImproves the text of machine-readable zones on passports, identity cards, and visas, and flags unconfirmed readings. (L#NAVI-189)
- ChangedChanges the default OCR language to automatic detection, so each document is read in its own language. (L#NAVI-115)
- ChangedExports the charts embedded in an Excel workbook as data tables when converting it to Markdown or IR Lite. (#3826)
- ChangedMarkdown export now applies the lexicon pack of the configured OCR language (
OcrSettings.DefaultLanguages) to born-digital text, whose words carry no language of their own, andOcrSettings.DetectLanguageFromDigitalText(advanced, off by default) identifies born-digital pages from their text without an OCR pass when an explicit language is configured. (L#NAT-633) - ChangedChanges layout reconstruction to require the VLM-enhanced ICR license, as the VLM setting does. (#3607)
- ChangedChanges VLM-enhanced recognition to keep OCR running by default, so text outside the detected regions and table content is no longer missing from scanned pages. Where the VLM’s transcription of a region is kept, the region’s words spell that transcription; a word OCR didn’t find gets an estimated position, marked
boundsEstimatedin the JSON output. Set the OCR activation mode to deactivated to make the VLM the only text source. (#3446) - ChangedChanges document classification and splitting to read page text with
AdaptiveOcrunless a vision engine is set. (#3809) - ChangedChanges the
Vision.Decidecontext window from 8K to 32K tokens, which uses about 0.6 GB more memory per process. (#3698) - ChangedChanges
Vision.Decideto answer requests it can’t accept with a 400 and codeinvalid_inputinstead of a 500. (#3728) - ChangedChanges
Vision.Decideto fail when the decision model calibration is unavailable instead of returning uncalibrated probabilities. (#3708) - FixedFixes layout reconstruction not running on scanned pages, now with the local vision model too. (#3607)
- FixedFixes lower OCR accuracy on GPU machines when several pages or documents share the GPU. A time limit canceled the GPU text-recognition pass while it waited for the GPU, so those pages lost that pass. The limit is now off by default, and a recognition pass that is cut short or fails is reported. (L#NAVI-127)
- FixedFixes words read twice on scanned pages when the recognition pipeline runs directly on a GPU machine. (#3607)
- FixedFixes a repaired word in a PDF text layer picking up text from the line above or below it. (#3620)
- FixedFixes a value printed beside its label being read before the label. (#3609)
- FixedFixes Auto-mode extraction failing with an internal error on documents whose pages all route to the structure tier. (#3541)
- FixedFixes Auto-mode pages routed to the understand and agentic tiers returning corrupted text on GPU machines: They now run the same evidence-fusion OCR pipeline the explicit modes use. (#3562)
- FixedFixes barcode blocks in parsed documents covering the whole surrounding picture instead of only the barcode. (L#DWS-853)
- FixedFixes document decision classification falling back to CPU when it’s the first operation on a GPU host. (#3800)
- FixedFixes duplicate text recognition when printed and handwritten regions partially overlap. (L#NAVI-69)
- FixedFixes extraction losing the text of a zone printed on the same line as another, such as a value beside its label. (#3609)
- FixedFixes form values and check marks that no detected region claimed being output in one block, away from their labels; each now follows the text beside or above it in the reading order. (#3611)
- FixedFixes grounding scores missing from the confidence results of windowed and schema-split extractions. (#3647)
- FixedFixes invoice headers whose multiline addresses were split into several table rows. (L#NAVI-134)
- FixedFixes label and value lines printed above a table being merged into the table. (L#NAVI-180)
- FixedFixes Markdown export fusing a hyphenated compound whose own hyphen sits at a line wrap, so quasi-experimental no longer becomes quasiexperimental. (L#NAT-625)
- FixedFixes missing font and bold or italic styles on words that Vision recognizes late on scanned pages, such as table cells, and missing font sizes on scanned images that declare their resolution. (#3756)
- FixedFixes ordinary words being reported as superscript or subscript on gray and low-contrast scans. (#3449)
- FixedFixes paragraphs that share a reading position appearing in a different order from run to run in the text layer of OCR’d PDFs. (#3616)
- FixedFixes reading order on pages where a value is printed beside its label, such as an email header or a totals box: Each label now reads with its value before the text below it. (#3621)
- FixedFixes scanned zones losing their text when the page’s language couldn’t be detected. (#3618)
- FixedFixes small tables with one or two rows of values being exported as text instead of a table. (L#NAVI-177)
- FixedFixes some QR codes and barcodes being returned as pictures instead of their decoded value. (#3475)
- FixedFixes Specialist form field VLM calls using best-effort (not strict) structured output, which could silently degrade labeling accuracy by letting malformed fields through instead of being rejected up front. (#3652)
- FixedFixes superscript formatting in Markdown exports when inline formatting is enabled. (L#NAVI-195)
- FixedFixes table cells that held the values of two neighboring columns. (L#NAVI-178)
- FixedFixes table cells that wrap over two lines being split into two rows. (L#NAVI-134)
- FixedFixes table header rows being exported as a heading above the table. (L#NAVI-179)
- FixedFixes tables whose first columns were exported as text beside the rest of the table. (L#NAVI-180)
- FixedFixes tables whose total line was merged into the last row. (L#NAVI-176)
- FixedFixes text extraction on rotated PDF pages. (L#NAVI-111)
- FixedFixes text painted in the same color as its background appearing in ICR output. (#3625)
- FixedFixes text printed twice when one recognized word covered both a label and its value, such as a tax number after its label. (#3617)
- FixedFixes the Specialist form field VLM creating fields not grounded in any real detection on the page. (#3669)
- FixedFixes checkbox and radio button states being misread and requires non-empty field labels and descriptions during AI form field extraction. (#3652)
- FixedFixes Vision coordinates that could differ slightly between two GPU runs on the same document. (L#NAVI-192)
- FixedFixes Vision extraction results that could change when the same document is processed again. (L#NAVI-192)
- FixedFixes Vision warmup skipping the local vision model that the Auto engine and layout reconstruction use. (#3614)
- FixedFixes
Vision.Decideanswering a question with empty instructions with a 500 instead of a 400invalid_input. (#3753) - FixedFixes words of two text lines read interleaved, and in a different order from run to run, when a tall graphic on the same line was recognized as text. (#3615)
2.1.12 - 4 Sep 2026
General
- AddedAdds an optional Fast mode for lower-latency OpenAI processing.
- AddedAdds zero-configuration local vision processing with automatic model downloads and GPU acceleration. (J#NAT-551)
- ChangedImproves born-digital text extraction by detecting citation markers, footnote numbers, and bracketed references as super/subscript and promoting heading candidates found in the text layer to heading zones.
- ChangedImproves the agentic vision tier by reporting a structural role for each text zone and promoting high-confidence
TitleandSectionHeaderdetector Text zones to headings. - ChangedImproves vision layout on born-digital pages by separating larger or bold heading lines inside detected text zones into their own heading zones. This behavior is controlled by
splitHeadingsInTextZones, which is on by default. - ChangedImproves chart data extraction on multipanel and multilevel charts by asking the model for one table per panel, including the panel title in a
<caption>, and folding extra grouping levels into the row label. - ChangedChanges the default OpenAI vision model to
gpt-5.6-luna. - FixedFixes an issue where vision extraction failed with
Failed to crop image to zone boundson born-digital pages whose ruled frame extended past the page box. It also prevents decorative horizontal rules from being admitted as page-wide tables. - FixedFixes an issue where
License.TraceFeaturesreturned an empty or single-entry list in trial mode. It now reports every licensed component the application used.
OCR
- AddedAdds form field fill-state detection so detected form fields now report whether they’re filled or empty, including checked/unchecked checkboxes, signed/blank signatures, and filled/empty text fields.
- AddedAdds checkbox and form field state to Markdown export. Checkboxes render as task-list items, and labeled form fields render as
label: valuelines. - AddedAdds the
Ocr.detectLanguageOncePerDocumentsetting, which controls whether automatic language detection runs once for the whole document or independently on every page. - AddedAdds a language tag to Markdown code fences, inferred from the code block’s own content.
- AddedAdds document splitting, which finds sub-document boundaries in a merged document and returns the page ranges.
- ChangedImproves Markdown export of text that wraps across lines. A word broken by a hyphen at the end of a line is rejoined into one word, while a wrapped number range keeps its hyphen.
- ChangedSpeeds up automatic OCR language detection by identifying the language from a band of the page instead of the whole page. This behavior is controlled by the new
Ocr.detectLanguagePageFractionsetting. - ChangedSpeeds up automatic OCR language detection by running its text-recognition pass for speed rather than accuracy. This behavior is controlled by the new
Ocr.detectLanguageFavorAccuracysetting. - ChangedChanges zero-shot classification so single-label and multilabel modes may now return different predicted labels and rankings.
- ChangedChanges table export so a header row detected by the table model is emitted with header cells instead of plain cells.
- ChangedChanges the classification confidence threshold to apply on each mode’s own confidence scale, so single-label and multilabel documents may abstain differently.
- ChangedChanges Markdown export to apply inline bold and italic styling to headings and page headers and footers, and to emit runs that are both bold and italic as
**_text_**rather than***text***. - ChangedChanges Markdown export to include page headers and footers by default, fenced by page-header and page-footer comment markers.
- FixedFixes an issue where Markdown export dropped every chart table after the first when a chart held several panels.
- FixedFixes an issue where table cell column and row spans reached past the end of the table, which produced malformed tables in the Markdown and JSON exports.
- FixedFixes an issue where table extraction on tall tables whose recognized structure collapsed to a couple of rows lost rows. The table region is now reread in horizontal bands, and the lost rows are recovered.
- AddedAdds bold detection from font weight metadata in born-digital PDF text extraction.
- ChangedImproves text style extraction by detecting underlines in born-digital PDFs and resolving bold and italic styles from font descriptors.
Vision
- AddedAdds a reasoning-effort setting for AI data extraction and vision zone classification, trading model reasoning time against cost.
- ChangedImproves chart detection when extracting document content with vision language models.
- ChangedImproves NLI grounding confidence on numeric table cells by anchoring the verbalized claim to the cell’s printed value and identifying its row by all its columns, so correct numeric extractions no longer receive low groundedness.
- ChangedChanges AI data extraction to enforce the requested JSON schema on the provider and fall back automatically for models that don’t support it.
2.1.11 - 21 Aug 2026
General
- ChangedImproves exception-to-status mappings and adds
DocumentConversionFailedandVisionOperationFailedstatus codes. (J#NAT-603) - ChangedImproves first-request latency on GPU deployments by preparing the CUDA runtime during warmup instead of during the first request.
- ChangedImproves multipage CPU processing throughput by scheduling concurrent page inference on a shared thread pool.
- FixedFixes GPU memory growing across requests and never being released when vision inference runs on the CUDA provider.
OCR
- AddedAdds a maximum-threads OCR setting that processes pages concurrently when making a document searchable.
- AddedAdds offline language detection for Tamil, Telugu, Kannada, Malayalam, and Bengali, with faster script models.
- ChangedImproves OCR speed on modern x86 processors by using VNNI instructions when available.
- ChangedOptimizes automatic OCR language detection: detects once per document and skips it when page OCR is disabled.
- FixedFixes the OCR accuracy-vs.-speed setting being ignored; disabling accuracy mode now selects the faster recognition path.
- AddedAdds a PDF validation API that verifies PDF/A and PDF/UA conformance and produces detailed validation reports.
- AddedAdds dedicated OCR and PDF editing exception types that identify the failed operation and page.
- AddedAdds the
AutoTaggingEngineaccessibility setting to choose betweenHeuristic,Vision, andVisionVlmtagging. - AddedAdds vision-backed layout detection to PDF/UA auto-tagging for higher-fidelity document structure.
- FixedFixes
MakeSearchablewriting every page’s OCR text layer onto the first page of multipage documents.
2.1.10 - 10 Aug 2026
General
- AddedAdds GPU-accelerated vision inference on Linux via the WebGPU execution provider on Vulkan-capable GPUs. (J#NAT-560)
- AddedAdds Windows GPU-accelerated vision inference via the WebGPU execution provider on Direct3D 12. (J#NAT-560)
- AddedAdds a local in-process vision language model provider for AI document augmentation without a cloud or server dependency. (J#NAT-521)
- AddedAdds a natural language inference (NLI) groundedness confidence score and an opt-in v2
result.jsonmetadata schema for AI document extraction. - AddedAdds optional document restoration that enhances extraction for perspective-distorted document photos before extraction. (#2863)
- AddedAdds support for newer Anthropic models in the Vision API. (J#NAVI-32)
- AddedAdds support for opening password-protected PDF documents by supplying the password through the document open settings.
- ChangedRenames Claude vision provider APIs to Anthropic. (J#NAVI-31)
- FixedFixes Java SDK stream methods so documents can be opened from
FileInputStreamand exported toFileOutputStream. (J#NAT-106) - FixedFixes a process-exit crash when GPU vision inference is requested on Linux hosts without a Vulkan-capable GPU. (J#NAT-579)
- FixedFixes document classification reporting only one processed page for multipage documents.
- FixedFixes error handling to throw proper exceptions for file-not-found and invalid handle states.
- FixedFixes nullable numeric getters returning errors for unset values in the Java and Python SDKs. (#NATL-586)
OCR
- FixedFixes
AdaptiveOcrextraction failing on PDF pages whose embedded text layer contains no extractable words.
- AddedAdds page-level automatic deskewing to the PDF editor.
Templating
- AddedAdds PowerPoint and Excel template-to-PDF/UA generation via new presentation and spreadsheet editors. (J#NAT-559)
2.1.9 - 10 Jul 2026
General
- AddedAdds
Vision.GenerateSchemafor generating extraction-ready JSON Schemas with optional cross-field constraint rules. - AddedAdds a diagnostics artifact to vision document extraction containing VLM token usage and provider context when
includeDiagnosticsis enabled. (J#NAT-501) - AddedAdds automatic orientation correction that straightens pages rotated 90/180/270 degrees before vision extraction. (J#NAVI-18)
- AddedAdds barcode recognition for 1D, QR, Data Matrix, PDF417, Aztec, Micro QR, and MaxiCode codes in the vision pipeline. (J#NAT-530)
- AddedAdds page selection to document extraction through a 1-based page-range setting such as
1-3,5. (J#NAT-464) - AddedAdds quad-point polygon geometry to document regions for accurate representation of rotated and skewed areas. (J#NAVI-23)
- AddedAdds ranked BM-25 text search over documents: Build a reusable index, and then query it for passages by page and line. (J#NAT-512)
- ChangedImproves formula recognition performance, roughly halving processing time on equation-heavy documents. (J#NAT-556)
- ChangedImproves table extraction speed by up to 6× on table-heavy documents. (J#NAT-524)
- ChangedChanges vision licensing checks to exactly match the capabilities each API call invokes.
- ChangedDisables local handwriting OCR by default, with an advanced setting to opt back in. (#2746)
- ChangedOptimizes vision document extraction with GPU acceleration on macOS, up to ~3× faster. (J#NAT-524, J#NAT-525, J#NAT-537, J#NAT-550, J#NAT-552, J#NAT-554, J#NAT-555, J#NAT-556)
- ChangedUpdates the default OpenAI and Anthropic vision models to
gpt-5.4andclaude-sonnet-4-6. (#2964) - FixedFixes AI extraction errors mislabeled as an unreachable endpoint when the model returns no content.
- FixedFixes Markdown table-cell export replacing en/em dashes and minus signs with hyphens and leaving formatting noise.
- FixedFixes a crash when processing documents concurrently with GPU-accelerated vision. (J#NAT-556)
- FixedFixes a memory leak where vision objects weren’t released after use, growing native memory over time. (J#NAT-515)
- FixedFixes an issue where the structured JSON export flattened multiline zones into a single line; zone text now preserves visual line breaks as newlines. (#2771)
- FixedFixes duplicate paragraph text appearing alongside extracted table content when table extraction is enabled. (#2953)
- FixedFixes token usage counts to include Gemini thinking tokens and OpenAI cached input tokens.
OCR
- AddedAdds fully offline language and text-direction detection for documents, with a text-only mode that needs no document.
- AddedAdds multilingual zero-shot document classification, automatically handling non-English documents and labels.
- AddedAdds text-only zero-shot classification for text you already have, with no document opened or rendered.
- AddedAdds zero-shot document classification scoring text and images against caller-supplied labels.
- ChangedImproves concurrent processing of vision documents. (J#NAVI-21)
- ChangedImproves offline language detection by suppressing spurious secondary languages, via new OCR settings.
- FixedFixes a race condition that could corrupt results when running vision OCR extraction concurrently. (J#NAVI-19)
- FixedFixes data tables being misclassified as a table of contents and dropped during vision data extraction.
2.1.8 - 13 Jun 2026
General
- ChangedImproves classification for handwriting zones for the VLM pipeline. (#2737)
- ChangedImproves the Java API reference layout with per-tag folder hierarchies for settings classes. (J#NAT-481)
- ChangedChanges the OCR setting to support multiple modes, including an automatic mode that selects the best processing pipeline. (#2738)
Office
- AddedAdds rendering of
DisplayBarCodeQR code fields in Word documents during conversion. (J#MAMBO-2058) - AddedAdds rendering of form checkboxes and dropdowns from ODT documents during conversion. (J#MAMBO-2072)
- FixedFixes a crash when converting Word documents containing an IF mail-merge field that resolves to an empty result. (J#MAMBO-2076)
- FixedFixes DOCX repair so footers stored with an invalid
hdrroot element are corrected toftr. (J#MAMBO-2078) - FixedFixes an invalid image stream exception. (J#MAMBO-2077)
- FixedFixes MSG files with custom
IPM.Notesubclasses, such as archived emails, being reported as an unknown format. (J#GDPIC-1621)
- ChangedImproves embedded font subsetting with corrected
.notdefglyph handling and CFF font reconstruction. - ChangedChanges the Producer and Creator metadata of generated PDF documents from
GdPicture.NETtoNutrient SDK. (J#NAT-460) - FixedFixes PDF/A validation errors related to embedded font glyph and width consistency. (J#PDFA-95)
2.1.7 - 8 Jun 2026
General
- AddedAdds a text export format for converting documents to plain text.
- AddedAdds an opt-in strict structured-output mode for AI data extraction with automatic schema normalization.
- AddedAdds first-class form-field detection to the Vision API, with optional AI semantic labeling of detected fields.
- AddedAdds intelligent chart parsing via HTML tables. (J#NAT-457)
- AddedAdds PNG, JPEG, and BMP support to image export, with automatic format detection from the destination filename. (J#NAT-474)
- AddedAdds single-flight coordination in
ResourceDownloaderto coalesce concurrent requests for the same model download. (J#NAT-472) - AddedAdds support for primitive, enum, and string-array properties in settings. (J#NAT-107)
- AddedAdds
Vision.ExtractStructuredfor AI-powered structured data extraction shaped by a caller-supplied JSON Schema. - ChangedImproves AI extraction robustness with fail-fast schema validation and culture-independent confidence scoring.
- ChangedImproves AI model download reliability with integrity validation and automatic retry on failure. (J#NAT-419)
- ChangedImproves form-field detection by removing overlapping predictions for cleaner, more consistent results. (J#NAT-471)
- ChangedImproves heuristic PDF layout extraction.
- ChangedImproves the Office-to-PDF speed tests. (J#MAMBO-2073)
- RemovedRemoves form-field inference from vision intelligent content recognition. (J#NAT-469)
- FixedFixes intermittent errors in the
VlmEnhancedvision pipeline when processing documents concurrently. (J#NAVI-16) - FixedFixes the
GdPicture.NET.ToolkitNuGet package on .NET Framework 4.6.2 by including native runtimes as a package dependency. (J#GDPIC-1619) - FixedFixes incorrect text appearing inside images in multipage document Markdown output. (#2654)
- FixedFixes loss of image alternate text descriptions when converting documents to PDF/UA. (J#PDFA-93)
- FixedFixes word alignment in document layout that caused paragraphs to absorb text from neighboring lines. (#2633)
Office
- ChangedImproves DOCX line spacing rendering for documents using the legacy line spacing compatibility option. (J#MAMBO-2027)
- ChangedImproves ODT-to-PDF page dimension rendering to match Microsoft Word. (J#MAMBO-2059)
- ChangedImproves parsing of empty spaces from ODT documents. (J#MAMBO-2075)
- FixedFixes an exception when loading a Word document that contains an empty custom XML part. (J#MAMBO-2067)
- FixedFixes an exception when rendering table cells where all text falls outside the cell boundary. (J#MAMBO-2065)
- FixedFixes missing spacing before a numbered paragraph when the preceding paragraph isn’t part of the list. (J#MAMBO-2062)
- FixedFixes ODT tables losing their header row. (J#MAMBO-2074)
- AddedAdds support for converting between Instant JSON and GdPicture XMP annotations. (J#NAT-406)
- AddedAdds support for exporting Instant JSON. (J#NAT-397)
- ChangedImproves PDF-to-Markdown conversion performance.
- ChangedEnables change tracking for Instant JSON exports by default. (J#NAT-394)
- FixedFixes empty embedded file handling when merging PDF documents. (J#GDPDF-1520)
Templating
- AddedAdds PowerPoint (PPTX) templating with placeholders, loops, conditionals, tables, and images to the Office templater. (J#OFTAPI-20)
2.1.6 - 25 May 2026
- AddedAdds the
AdaptiveOcrvision engine — a heuristic-first PDF extraction pipeline with automatic OCR fallback for image-based pages. - AddedAdds a born-digital text fast path to VLM pipelines and enables VLM-only text output when OCR is disabled. (#2607)
- AddedAdds direct PDF processing support to the Vision API for improved document analysis.
- AddedAdds support for creating searchable PDFs.
- AddedAdds form detection support.
- ChangedImproves stability when processing multiple documents in a single process. (J#NAVI-8)
- ChangedImproves vision data extraction fidelity.
- FixedFixes caption and footnote relationships being linked to incorrect element IDs. (#2615)
- FixedFixes a crash when running OCR on Linux. (J#NAT-458)
- ChangedImproves Java SDK conversion, rendering, and extraction capabilities by incorporating enhancements from the Nutrient .NET SDK 14.4.3(opens in a new tab) release line.
2.1.5 - 8 May 2026
- AddedAdds Markdown export format to Vision API content extraction, including options for semantic block formatting and header/footer inclusion. (J#NAT-347 and #2590)
- AddedAdds a failure message when
VlmEnhancedIcrisn’t configured correctly. (#2534) - ChangedImproves Java API reference documentation with class declarations, inheritors, settings access patterns, and tag-grouped navigation. (J#NAJA-7)
- ChangedImproves Vision API JSON export performance on large multipage documents. (#2499)
- ChangedImproves Java SDK conversion, rendering, and extraction capabilities by incorporating enhancements from the Nutrient .NET SDK 14.4.2(opens in a new tab) release line.
2.1.4 - 12 Apr 2026
- ChangedImproves Java SDK conversion, rendering, and extraction capabilities by incorporating enhancements from the Nutrient .NET SDK 14.4.1(opens in a new tab) release line.
- ChangedImproves handwriting vision capabilities by adding a new in-house model that better identifies handwritten regions in images.
2.1.2 - 27 Feb 2026
- FixedFixes XLSX conversion and OCR memory usage issues as part of a coordinated .NET/Java/Python SDK release. (#2309)
- ChangedImproves the Vision API for ICR and fixes issues in the VLM pipeline. (#2309)
2.1.1 - 30 Jan 2026
- AddedAdds annotation editing, forms editing, redaction, and signature support to improve feature parity with Nutrient .NET SDK(opens in a new tab). (#2302)
- ChangedExpands customizable settings support. (#2302)
- ChangedImproves vision capabilities, including OCR table extraction, intelligent content recognition (ICR), vision language model (VLM)-augmented ICR (local VLM/Claude/OpenAI), image description, and automatic download of encrypted models. (#2302)
2.1.0 - 29 Jan 2026
- AddedAdds PDF page and page-collection editing capabilities. (#2175, #2251, #2286, #2289)
- AddedAdds form field creation capabilities for text, checkbox, radio button, push button, combo box, list box, and signature fields. (#2175, #2251, #2286, #2289)
- ChangedChanges Java SDK capabilities as part of the 2.1.0 milestone. (#2175, #2251, #2286, #2289)
2.0.0 - 18 Jul 2025 and 21 Jul 2025
- AddedAdds the first public release of the Java SDK 2.x line on Maven Central, separate from the previous 1.x branch. (#1931, #1933, #1934, #1940, #1943, #1946, #1949, #1959, #1970, #1973, #1977, #1978)
- ChangedChanges the SDK architecture and simplifies the API. (#1931, #1933, #1934, #1940, #1943, #1946, #1949, #1959, #1970, #1973, #1977, #1978)
- ChangedImproves handling of complex parameters and exceptions for Java developers. (#1931, #1933, #1934, #1940, #1943, #1946, #1949, #1959, #1970, #1973, #1977, #1978)