docling Releases

Introduces performance tracking for VLM inference time and configuration options for ONNX Runtime graph optimization.

Mar 6, 2026

Adds WebVTT export format and fixes Excel image handling, OCR configuration, and Unicode normalization issues.

Mar 2, 2026

This version introduces dynamic actions for changing `--with-nth` and `--header-lines`, alongside performance optimizations ranging from 1.3x to 1.9x faster filtering.

Mar 2, 2026

This release introduces XBRL instance report parsing support and unifies model-family inference engines with KServe v2 API support. It also includes fixes for ASR segment handling and DOCX hyperlink processing.

Feb 24, 2026

This release introduces docling-parse v5 while deprecating old backends, along with critical security fixes for XML External Entity vulnerabilities. It also improves CSV handling and engine options deserialization.

Feb 17, 2026

This is a bugfix release addressing ASCIIDoc image handling, timezone-aware datetime conversion, and failed page tracking in documents.

Feb 13, 2026

This release adds LaTeX document parsing support, introduces a pluggable VLM runtime system, and provides inference engines abstraction for object detection models using HF Transformers and ONNX runtime.

Feb 11, 2026

Version 1.1, titled 'Voyager,' introduces significant AI enhancements including multi-pass refinement and AI regeneration for resume sections, alongside UI improvements like draggable elements and inline editing of titles. The release also improves platform compatibility, specifically fixing Windows bugs and Docker configurations, and expands internationalization support to include Portuguese and several Asian languages.

Feb 10, 2026

This release introduces chart extraction models and improves Excel table detection with merged cell handling. It also addresses PowerPoint image reference issues.

Feb 3, 2026

Adds WebVTT source tracking and Word document comments extraction capabilities.

Jan 30, 2026

Removes support for Python 3.9 and fixes Markdown pipe symbol handling and PPTX bullet point grouping.

Jan 23, 2026

Fixes a page indexing off-by-one error in the VLM pipeline.

Jan 21, 2026

This release introduces a new picture classifier v2.0 and adds classification filters for picture descriptions. It also includes various fixes for XPU compatibility, logging standardization, and dependency version updates.

Jan 20, 2026

The update adds support for DeepSeek-OCR in the VLM pipeline and includes fixes for logging, DOCX grouped pictures, and documentation typos. It also adds Semantica integration.

Jan 13, 2026

This release focuses on enrichment annotations in the new meta format, adds XPU device support for Intel GPUs, and introduces an option to report timing details. It also includes fixes for dependencies, type hints, and Transformers models.

Jan 9, 2026

This is the first stable release (v1.0.0) of Resume Matcher, marking a milestone with the introduction of core capabilities such as tailored resume creation, multiple ATS-compatible templates, and support for multiple AI providers. The release also includes easy-to-use Docker support and lays the foundation for the application's stability and feature set.

Jan 6, 2026

The release adds a preset for using granite-docling via vLLM and other APIs. It includes fixes for DOCX tables with merged cells and markdown parsing, along with documentation updates for RTX GPU performance.

Dec 24, 2025

This version adds a YAML output format to the CLI and fixes issues with RapidOCR parameter names and DOCX paragraph styles. It also includes documentation for PipelineOptions and GPU benchmarks.

Dec 15, 2025

A maintenance release addressing various bugs including word/char cells in full page OCR, missing fonts in RapidOCR artifacts, image scale generation, and HTML/DOCX parsing issues.

Dec 9, 2025

Firecrawl v2.7.0 introduces enterprise search capabilities, enhanced branding extraction, and a new NOQ scrape system. The release includes significant SDK enhancements for JavaScript and Python, improved error handling, and performance optimizations for self-hosted deployments.

Dec 5, 2025