aifollow.news Search
← TrendingLatest development Oct 28 00:41
STORY IN FOCUS

SoMark releases OXR-1.0 weights and offers a free document parsing API

SoMark has released the weights for OXR-1.0, a 1.3 billion parameter document parsing model, and offers a free API that requires no account or key. It converts PDFs and page images to Markdown or JSON and handles page elements including tables and formulas. Its training loop generates new data from parsing errors, but can only address errors the assisting vision language model detects.

1 developments3 reports2 sources

Event developments

Coverage of the same occurrence shares a node; subsequent developments have their own nodes.

SoMark releases OXR-1.0 document parsing weights and a free API

OXR-1.0 turns PDFs and page images into Markdown or JSON, identifying content such as tables and formulas and setting its reading order. SoMark says it ranks first on three document parsing benchmarks. Its weights are available on Hugging Face, and the free API needs no account or key; light testing on a local GPU requires 24GB of memory.

View source reports · 2 sources · 3 reports
All reports · 3

By original publication time, with each report's bookmarks and feedback preserved.

Rohan Paul (@rohanpaul_ai) Machine translation

SoMark releases OXR-1.0 weights and offers a free document parsing API

SoMark has released the weights for OXR-1.0, a 1.3 billion parameter document parsing model, and offers a free API that requires no account or key. It converts PDFs and page images to Markdown or JSON and handles page elements including tables and formulas. Its training loop generates new data from parsing errors, but can only address errors the assisting vision language model detects.

Rohan Paul (@rohanpaul_ai) Machine translation

SoMark releases OXR-1.0 document parsing weights and a free API

OXR-1.0 turns PDFs and page images into Markdown or JSON, identifying content such as tables and formulas and setting its reading order. SoMark says it ranks first on three document parsing benchmarks. Its weights are available on Hugging Face, and the free API needs no account or key; light testing on a local GPU requires 24GB of memory.

量子位 AI Machine translation

SoMark team releases OXR document parsing model with an online demo and free API

OXR recognizes 13 types of document elements, including tables and formulas, and converts PDFs and document images into Markdown or JSON while preserving reading order and layout coordinates. In results published by the SoMark team, OXR ranked first on three document parsing benchmarks. Its code and model weights are available, along with an online demo and free API for developers to try.