Reducto's r-1 parses documents in a single pass at 1 cent per page

Reducto announced r-1, the first model in a new parsing family built on a rewritten architecture. It replaces the company's multi stage agentic OCR with a single full page pass. Reducto says r-1 is more accurate than its most powerful legacy agentic models, faster, and up to 6x cheaper.

The model is available today in preview. It runs through Reducto's hosted Parse API on V3 and is enabled with a configuration flag. There are no open weights and no local checkpoint to self host.

What the single pass covers

Legacy Parse ran OCR, layout detection and post processing as separate stages, with optional vision language passes layered on top, and every extra model call added latency. r-1 folds text, tables, figures, layout, reading order, formatting and grounding into one pass. Each returned block carries page relative bounding boxes that tie content back to its position on the page.

According to the documentation, the pass natively handles digital text, scans and handwriting; table structure including merged cells and nested headers; columns, headers, footers, sidebars and reading order; figure detection with a short generated description; and formatting that carries meaning, such as headings, lists, bold, underlines and strikethroughs.

Pricing and the fine print

Legacy agentic models cost between 3 and 6 cents per page depending on workload. r-1 is 1 cent per page all in, with no feature multipliers or credit costs added to reach high accuracy.

Reducto reports a 20% reduction in error rate for the preview, but that figure is measured against its own prior pipeline rather than a third party baseline. The company also claims r-1 outperformed widely used hyperscaler products, naming Amazon Textract and Azure Document Intelligence, along with large LLMs on complex documents. Those comparisons are vendor run, and no public eval harness or dataset was released with the announcement.

Agentic processing has not disappeared. Pages that need custom prompts or advanced chart extraction still route through the agentic pipeline, which adds latency. Reducto has flagged two more pieces ahead: r-1 mini for speed and cost sensitive workloads, and automatic per page routing. Organisations switching from another parser can request up to $5,000 in credits for a side by side comparison.

More detail is in the MarkTechPost report.