Reducto Unveils a Frontier Parsing Model That Makes the World's Hardest Documents AI-Ready for 1¢ a Page

The new r-1 model is the first in a family of document intelligence models that reduces parsing errors by up to 20% while replacing complex, multi-tool pipelines with a single model that costs up to 6× less.

SAN FRANCISCO, Sept. 1, 2026 /PRNewswire/ -- Reducto today introduced r-1, a frontier document parsing model that turns complex PDFs, scans, spreadsheets, and other files into accurate, structured data for AI systems.

Reducto has already established itself as a leader in document parsing by handling the long tail of complex files that break conventional tools. Available today in preview, r-1 extends that lead—reducing errors by up to 20%, improving latency at high volumes, and bringing the all-in cost down to 1¢ per page.

Before an AI system can reason over a document, it must first understand what is on the page. That becomes difficult when meaning is encoded not only in text, but also in tables, handwriting, reading order, formatting, checkboxes, strikethroughs, and the position of content. A parser that misreads a financial table can give an AI agent the wrong numbers. One that drops a strikethrough can reverse the meaning of a contract.

Established services such as Amazon Textract and Azure Document Intelligence helped make cloud document processing widely available. But organizations working with complex documents often still combine multiple tools, models, and layers of post-processing to achieve the accuracy they need. In Reducto's evaluations, r-1 outperformed commonly used hyper-scaler products and large LLMs on complex documents while providing a complete parse at one all-in price.

Built from the ground up and trained on some of the most challenging document data in the world, r-1 combines layout detection, reading order, tables, formatting, grounding, and granular citations in a single model. It is designed for the long tail of documents where simpler parsers break down, including dense tables, unusual layouts, low-quality scans, handwriting, watermarked content, and files that do not follow predictable templates.

By combining these capabilities, r-1 can handle an organization's full document workload without requiring teams to route files among providers or maintain separate pipelines for different document types. Its flat price of 1¢ per page includes the complete parse, without additional model charges or feature-based multipliers.

r-1 is the first in a broader family of Reducto parsing models designed for different points on the accuracy, latency, and cost curve. Planned additions include r-1 mini, a smaller model for speed- and cost-sensitive workloads, and automatic routing that selects the right model for each page.

Organizations using another parser can receive up to $5,000 in credits to test r-1 on their most difficult documents, along with hands-on benchmarking support, at reducto.ai/migrate. r-1 is also available today in preview through a configuration flag in the Reducto Parse API.

About Reducto

Reducto is an agentic document platform that turns complex, real-world documents into structured data for AI agents and document-intensive workflows. The company has raised more than $108 million from investors including Andreessen Horowitz and First Round Capital and has processed more than a billion pages a month. Its customers include Harvey, Scale AI, Vanta, Airtable, Toast, and Fortune 10 enterprises.

SOURCE Reducto