AI Briefing
KO

OvisOCR2 Model Released

·2026.07.13 19:55

Key point

A lightweight end-to-end document structuring OCR model based on Qwen3.5-0.8B has been released.

Details

Built on Qwen3.5-0.8B, OvisOCR2 is an end-to-end OCR model that directly converts document pages into Markdown format including text, tables, formulas, and reading order.

Key features are as follows:

  • Lightweight size: With 0.8B parameters, it enables efficient operation in local environments.
  • High performance: It recorded a score of 96.58 on OmniDocBench v1.6 and 75.06 on PureDocBench.
  • Development convenience: Weights are released under the Apache 2.0 license, and it supports vLLM.

This model is optimized for extracting text while preserving the structural information of documents.

ocrqwenmarkdowndocument-parsingopen-source-modelovisocr2
Read originalShare on Threads

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.

Related

RedditUkisAI Releases Swift Model Boosting Qwen 3.8 27B Inference Speed by 1.95xPyTorchKRTrillion Labs Releases GravityOCR, an OCR Model Applying Self-Speculative DecodingHacker NewsApple Releases LensVLM-9B for Processing Compressed Text ImagesTLDR AIQwen Releases 'RecreationWorld' for Training AI Agents to Rebuild AppsRedditSenseNova-U1.5-8B-MoT Released: Unified Text-to-Image and Understanding Model Initialized from Qwen3PyTorchKRvLLM Releases 'Decision 1.0', AI Models Specialized for Decision-Making