v1.14.2 v1.15.0
0xNyk/council-of-high-intelligenceAug 13, 2026 → Aug 17, 20264 days between releases
What's New
Adds selective OCR capabilities using PDFium and PP-OCRv6 Small for scanned and mixed PDFs.
New Features
- Selective OCR pipeline for scanned and mixed PDFs.
- OCR provenance, confidence, and timing data.
- Offline operation capabilities.
- Reuses OCR sessions for faster extraction.
- Exposes OCR pipeline across Rust, pdf2md, Node, and Python.
Improvements
- Recovers credible PDFium text layers before OCR.
- Pinned runtime and model caching for stability.
- Preserves row boundaries during OCR extraction.
Upgrade Notes
Requires new model dependencies (PP-OCRv6 Small) and pinned runtime. Output format may change if OCR is enabled on previously unscanned documents. Ensure offline environment support if utilizing offline features.
The Two Releases
1.14.2 — Hardened parsing for pathological PDFs
Aug 13, 2026· 2 months ago
This release prioritizes security and stability by hardening parsing against crafted PDFs that could potentially cause denial-of-service or memory exhaustion.
Full release notes1.15.0 — Selective OCR for scanned and mixed PDFs
Aug 17, 2026· 2 months ago
This is a major feature release introducing selective OCR capabilities for scanned and mixed PDFs, utilizing PDFium rendering and PP-OCRv6 Small.
Full release notes