v2.133.0
docling-project/doclingv2.133.0Oct 3, 2026by docling-ops[bot]
Upgrading?
AI Summary
This release focuses on improving document parsing accuracy for various formats, specifically addressing issues with LaTeX decoding, OpenDocument charts, and Markdown formatting. It also introduces configuration options for OCR models.
Key Highlights
- Added support for configuring RapidOCR model size.
- Fixed reading chart object titles in OpenDocument files.
- Improved LaTeX decoding for Windows-1252 sources and TikZ dependencies.
- Enhanced Markdown handling to preserve code block indentation.
New Features
- Allow configuring RapidOCR model size
Full Release Notes
### Feature * **ocr:** Allow configuring RapidOCR model size ([#4500](https://github.com/docling-project/docling/issues/4500)) ([`a1c5ff2`](https://github.com/docling-project/docling/commit/a1c5ff2a8c5ab2bf87d7c1eb896b5bc351f51220)) ### Fix * **opendocument:** Read the chart object's own title ([#4463](https://github.com/docling-project/docling/issues/4463)) ([`4c109c2`](https://github.com/docling-project/docling/commit/4c109c255f3655af2c6f2c0be8c8677adee41a8a)) * **code-formula:** Keep the extracted text when a batch fails ([#4495](https://github.com/docling-project/docling/issues/4495)) ([`a25aa1d`](https://github.com/docling-project/docling/commit/a25aa1dedccf74572df0385cde0cf02fb059b92c)) * **latex:** Decode Windows-1252 sources and 8-bit TikZ dependencies ([#4475](https://github.com/docling-project/docling/issues/4475)) ([`b5aa0ef`](https://github.com/docling-project/docling/commit/b5aa0ef38aa5577b812496d70b1ca9f6d0142b00)) * **md:** Keep the first-line indentation of code blocks ([#4487](https://github.com/docling-project/docling/issues/4487)) ([`cc937ec`](https://github.com/docling-project/docling/commit/cc937ec6c07b2ae5a71b7e3b726cade46083086b)) * **html:** Keep dt/dd groups wrapped in div elements in description lists ([#4390](https://github.com/docling-project/docling/issues/4390)) ([`f1c42e3`](https://github.com/docling-project/docling/commit/f1c42e394e3f5c40375c83edf01f8de762bb64f9)) * **latex:** Decode \input files the same way as the main file ([#4441](https://github.com/docling-project/docling/issues/4441)) ([`bba2ec5`](https://github.com/docling-project/docling/commit/bba2ec58fe58af1be630fea5cc0d35b0ad2a472e)) * **docx:** Read w:numPr from the paragraph's own properties (#4456) ([#4457](https://github.com/docling-project/docling/issues/4457)) ([`1d4b9ae`](https://github.com/docling-project/docling/commit/1d4b9ae17b8a18ed723a0c9fc6e01bef080b3387)) * **csv:** Sniff dialect after leading empty lines ([#4467](https://github.com/docling-project/docling/issues/4467)) ([`2dd8c53`](https://github.com/docling-project/docling/commit/2dd8c533cdefeb77ccefaf3f47fe3be9714ff782)) * Detect the format of path inputs with an upper-case extension ([#4430](https://github.com/docling-project/docling/issues/4430)) ([`89935d5`](https://github.com/docling-project/docling/commit/89935d50fa378f1b32f576a40d9acbe7ab4e209e))
Related Repositories
public-apis/public-apis★ 491,128
A collective list of free APIs
NousResearch/hermes-agent★ 255,901
The agent that grows with you
yt-dlp/yt-dlp★ 200,002
A feature-rich command-line audio/video downloader
microsoft/markitdown★ 191,481
Python tool for converting files and office documents to Markdown.
Significant-Gravitas/AutoGPT★ 187,700
AutoGPT is the vision of accessible AI for everyone, to use and to build on. Our mission is to provide the tools, so that you can focus on what matters.