Skip to content
Trending storyDeveloping

ModelScope releases OvisOCR3, a unified 0.8B document-understanding model

1 article1 sourcesince Oct 28Last article 2h ago ·

Overview

AISummary of 1 article

ModelScope releases OvisOCR3, a 0.8B model that converts document images into structured data across five OCR and document-understanding tasks.

Task prompts select each capability, so one checkpoint replaces a separate model for each workflow, according to ModelScope's announcement on X. The model is released under the Apache-2.0 License.

ModelScope says OvisOCR3 leads comparable-size models in document parsing, general OCR, key information extraction, figure captioning, and cross-page table reconstruction. The team's claim is based on its own comparison, and no independent benchmark results are reported.

Written by AI from the articles below · updated Oct 28, 4:22 AM ET

Check the sources:

Latest developmentOct 28, 4:16 AM ET
OvisOCR3 is a unified 0.8B model for five document tasks

Article timeline

The articles in this story. Times are ET.

Oct 28
  1. ModelScopeOfficial
    OvisOCR3 is a unified 0.8B model for five document tasks

    AIModelScope releases OvisOCR3, a unified 0.8B model that turns document images into structured data across five OCR and document-understanding tasks. The team reports it leads comparable-size models in document parsing, general OCR, key information extraction, figure captioning, and cross-page table reconstruction. Task prompts select each capability, and the model is released under the Apache-2.0 License.

    Image from @ModelScope2022's post

Heat Trend

Not enough continuous observations to show a trend yet.