Summary
On September 1, 2026, Reducto introduced r-1, a document parsing model that handles complex layouts, tables, formatting, and citations in a single full-page pass rather than across multiple model calls, with bounding boxes linking each block to its position. Reducto says r-1 cuts errors up to 20% and brings all-in cost to about 1 cent per page, which it positions as up to six times cheaper than typical multi-tool pipelines.
What changed
Reducto shipped r-1 in preview, replacing multi-call parsing pipelines with a single-pass model that outputs structured text, tables, figures, layout, and positional bounding boxes.
Why it matters
Cheaper, more accurate single-pass parsing lowers the cost and latency of turning messy PDFs, scans, and spreadsheets into AI-ready structured data, a common bottleneck in RAG and agent pipelines. Collapsing a multi-tool pipeline into one model simplifies document ingestion for AI teams and undercuts incumbent parsing stacks on price.
Evidence excerpt
It delivers agentic-quality parsing for text, tables, figures, layout, and formatting in one full-page pass, with bounding boxes that link each block to its position on the page... all-in cost down to 1 cent per page.