Skip to content
AI Model Radar

aya vision 8b

CohereLabs/aya-vision-8b
Open Source · chat · open-weights
Open Alert me on changes
Context
16K
Max output
4K
Weights
Open
API $/1M
—
Modalities
text · image
Released
02 Mar 2025
License: cc-by-nc-4.0 · CohereLabs/aya-vision-8b
Download image Share on X Share on LinkedIn
AI summary
● machine-written

Aya Vision 8B: Open-weight multimodal model with 16K context

Aya Vision 8B is an 8-billion parameter vision-language model from Cohere supporting 23 languages, with input modalities for text and images and text output. The model has a 16K context window and 4K maximum output, designed for low-latency image and text understanding tasks including OCR and visual question answering. Released under CC-BY-NC-4.0 license as open-weight, it is available for local deployment or API access.

What's new
  • Open-weight model supporting 23 languages including English, French, Spanish, Arabic, Mandarin, Japanese, Korean, and others
  • Vision-language capabilities for image analysis, OCR, and visual question answering
  • 4K maximum output tokens with 16K context window
  • Multimodal input supporting both text and image modalities
Best for
Optical character recognition (OCR) tasksVisual question answering on imagesImage description and analysisMultilingual vision understanding
Sources

Source: https://huggingface.co/CohereLabs/aya-vision-8b