Skip to content

Aya Vision 8B

A multilingual vision-language model for OCR and QA

Visit official site

What Aya Vision 8B is

The model is described as a multilingual vision-language release from CohereForAI, built on C4AI Command R7B and a SigLIP2 vision encoder for OCR and visual reasoning.

What you can do with it

  • Run OCR and text extraction
  • Describe an image in words
  • Reason over visual content

Getting started

  • Install transformers from source
  • Load the model with AutoProcessor
  • Call generate to produce text output

Pros & cons

✓ What we like

  • Open weights are released for the research community
  • Image and text inputs produce text output

! What to watch out for

  • Use is subject to CC-BY-NC and the C4AI acceptable use policy
  • Loading requires installing transformers from source

FAQ

How many languages are supported?

The material states 23 languages, including Chinese, English and French.

What licence applies?

The material states CC-BY-NC along with the C4AI acceptable use policy.

Last reviewed: 2026-11-15

More AI image generation tools

View all →

How we review