South Korea3.67B parametersImage + text32K context

Explore Kanana 1.5 V 3B Instruct

Use Kakao's compact Korean-English vision-language model for image captioning, document understanding, OCR reasoning, and multimodal instruction following.

Model

kakaocorp/kanana-1.5-v-3b-instruct

Input

Configure your request

The image-to-text pipeline message array. Each user message may contain image items with a URL and text items; the model accepts text and one or more images and returns text.