LoginSign Up
← Back to all sparks
D

Docling

AI-ASSISTANTS
Velocity5.0

Document parsing and conversion toolkit for AI pipelines

docling-project.github.io/docling ↗

Docling 2.150 reads inline XBRL filings while patching formats daily.

document-parsingocrformat-coveragefinancial-filingsopen-source
◆Current state
Docling ships almost daily. 2.150.0 adds inline XBRL input with remote taxonomy lookup, so it can read the tagged financial statements companies file with regulators, and fixes email, ODF, JATS and image orientation handling. 2.149.0 added opt-in embedded PDF image extraction, a reworked CLI and mlx-audio speech recognition, and 2.148.0 added Surya v2.5 and an iCalendar backend.
◆Where it's heading
Docling keeps widening the inputs it accepts, now into structured financial filings, while adding OCR and speech model presets and fixing format edge cases. The pace is incremental; no single release changes the product's shape.
◆Prediction
Expect more domain formats and model presets in the 2.15x line, with fixes continuing to land almost every day.

◆Recent moves

  1. 4d ago

    Docling 2.150 reads inline XBRL financial filings

    Adds inline XBRL input with remote taxonomy resolution, plus fixes for email bodies, ODF table cells, JATS licence text and EXIF image orientation. Another new input format in a near-daily release run.

    View source ↗
  2. 6d ago

    Docling 2.149: embedded PDF images, new CLI, mlx-audio ASR

    Opt-in embedded PDF image extraction, a reorganised CLI, mlx-audio speech recognition and a MiniCPM-OCR preset, with about a dozen format fixes. It follows the pattern of the last month: one or two new model presets per release on top of format edge-case repairs.

    View source ↗
  3. 8d ago

    Surya v2.5, an iCalendar backend and PDF sub/superscripts

    Adds Surya v2.5, an iCalendar backend and PDF superscript and subscript detection, with fixes across many formats. It continues the pattern of new model presets plus wider format coverage.

    View source ↗
  4. 9d ago

    Inline whitespace fixes across backends

    Backends now follow a shared whitespace rule for inline text, plus fixes for AsciiDoc titles, LaTeX links and large VLM tables. Small correctness work between feature releases.

    View source ↗
  5. 11d ago

    New OCR presets: Unlimited-OCR 2, LightOnOCR-3, Nemotron 3.5

    Three new model presets, including grounded output for LightOnOCR-3 and a guarded multi-pass Nemotron 3.5 parser, continue Docling's role as a switchboard for document vision models. The rest is a long list of format fixes, from hanging Markdown parsing to images placed in Excel cells.

    View source ↗
  6. 15d ago

    RapidOCR page orientation and InputDocument.release()

    RapidOCR now detects page orientation and InputDocument.release() gives callers control over memory, with a long list of format fixes. Steady parsing quality work, including unloading page backends.

    View source ↗