Datagrunt 5.3: Image-Only Page Markers, Bundle Opt-Out Warnings & LLM Skip Integration
Datagrunt 5.3.0 introduces explicit image-only page markers across all OCR modes, bundle opt-out warnings for unwritten page images, bundle format 5.3, and automatic rationalization skips for image-only pages.
November 8, 2026
Datagrunt 5.2.2: Pure-Rust PDF Compute Kernels, Worker Pool Reuse & Offline Cloud Batch Mode
Datagrunt 5.2.2 delivers compiled pure-Rust PDF compute kernels with GIL detachment, Stage 1 worker pool reuse and image skipping, and offline cloud batch preparation and ingestion for 50% token cost savings.
November 7, 2026
Datagrunt 5.2.1: Robust Gemini Thinking Level Normalization & CodeQL Security Hardening
Datagrunt 5.2.1 brings flexible case-insensitive and whitespace-tolerant thinking_level configuration for Gemini models, alongside CodeQL regex range alert resolutions.
November 7, 2026
Datagrunt 5.2: Two-Stage Batch Corpus Extraction, Section Coalescing & Gemini Thinking Levels
Datagrunt 5.2.0 introduces two-stage batch PDF corpus extraction with pure-stdlib manifest loading, greedy section coalescing for LLM rationalization, and Gemini thinking_level controls.
November 7, 2026
Datagrunt 5.1: Schema-Driven PDF Extraction with Consensus & Grounding
Datagrunt 5.1.0 introduces rationalize(), extracting structured Pydantic records from PDF documents with multi-provider cross-family model consensus, quote and key grounding against visible elements, cross-page entity folding, and deterministic Polars exports.
November 6, 2026
Datagrunt 5.0.2: PyO3 0.30 & Rust Build Infrastructure Upgrades
Datagrunt 5.0.2 upgrades native Rust CPython bindings to PyO3 0.30.1, pins Maturin to 1.14.1, updates core dependencies fancy-regex and proptest, and refreshes dev toolchains.
November 4, 2026
Datagrunt 5.0.1: Faster Start-up
Datagrunt 5.0.1 loads each API only when you first use it, so import datagrunt and every PDF read start faster. No code changes are needed.
November 4, 2026
Datagrunt 5.0: A New PDF Engine
Datagrunt 5.0 replaces PDF extraction with a single PDFium-based pipeline: a typed document model in reading order, hidden-text flags, zero-write reading, atomic outputs and checksummed bundles. CSV, Excel and Parquet are unchanged.
November 4, 2026
Datagrunt 4.6.47: Internal Groundwork for Datagrunt 5.0
Datagrunt 4.6.47 has no user-facing changes. It carries internal groundwork for Datagrunt 5.0's new PDF extraction stack, which is not part of the public API yet.
November 4, 2026
Datagrunt 4.6.46: Internal Groundwork for Datagrunt 5.0
Datagrunt 4.6.46 has no user-facing changes. It carries internal groundwork for Datagrunt 5.0's new PDF extraction stack, which is not part of the public API yet.
November 3, 2026