Reference
Formats and components
Many input formats converge on one document model, which serializes to a handful of output formats. This page is the compatibility reference behind that claim.
What Docling reads
Documents
Page-based documents where layout carries meaning.
- DOCX
- DOC
- PPTX
- PPT
- ODT
- ODP
Web & publishing
Text and publishing formats with existing structure.
- Markdown
- Plain text
- HTML
- AsciiDoc
- LaTeX
- EPUB
- Box Note
Spreadsheets & data
Tabular and legacy data formats.
- XLSX
- XLS
- ODS
- CSV
- EBCDIC
Images
Scans and photographs, read through the OCR engine of your choice.
- PNG
- JPEG
- TIFF
- BMP
- WEBP
Audio & video
Recorded media, transcribed into the same document model.
- WAV
- MP3
- M4A
- AAC
- OGG
- FLAC
- MP4
- AVI
- MOV
- MKV
- WEBM
XML & Docling
Domain XML and native Docling representations.
- JATS XML
- USPTO XML
- XBRL XML
- METS/GBS
- Docling JSON
- DocLang
- DCLX
Messaging & captions
Email messages and timed text.
- EML
- MSG
- WebVTT
What Docling writes
Structured
Lossless representations of the full document model.
- JSON
- YAML
- DocLang
- DCLX
Markup & text
Human-readable output for downstream text pipelines.
- Markdown
- HTML
- Split-page HTML
- Text
- DocTags
- WebVTT
Retrieval
Serialized chunks ready for indexing and retrieval.
- Chunks (JSONL)
What Docling extracts
Page
- Image
- Number
- Header
- Footer
Text
- Header
- Paragraph
- List item
- Code
- Formula
Table
- Structure
- Cell
- Caption
Picture
- Image
- Class
- Description
- Caption
Document
- Reading order
- Chunks
- Bounding boxes
- Provenance