LandingAI launches enhanced document extraction system with new model family

LandingAI has launched Agentic Document Extraction (ADE) Gen2, a revamped document intelligence platform built around the DPT-3 model family. The update introduces a tree-based approach to document parsing, pricing based on character return rather than page count, and atomic grounding to specific lines or words. ADE Gen2 is now generally available, with free access for developers via the ADE playground and enterprise deployment options across major cloud providers and on-premises environments. The system features two models: DPT-3 Verity, which transcribes digital documents with bounding boxes and confidence scores, and DPT-3 Pro, which handles complex layouts, scanned pages, and non-Latin scripts. DPT-3 Verity is priced at 40% of DPT-3 Pro’s cost, with automated routing planned for fall 2026. Pricing models vary by service tier, with priority tiers charging higher rates for synchronous processing and standard tiers offering cost savings for asynchronous workflows. ADE Gen2’s Parse v2 response includes structured metadata, markdown output, and atomic grounding for precise document navigation. This enables features like PII redaction, document diffing, and reviewer UIs. LandingAI claims parsing costs as low as one cent per page with Verity on standard tiers, though actual costs depend on document complexity.