Kakao open-sources Kanana-2-V-Decide decision model on Hugging Face
The multimodal model returns probabilities, choices and scores instead of sentences and allows commercial use under the Kanana Open License.
Kakao released its multimodal decision model Kanana-2-V-Decide as open source on Hugging Face on November 6, 2026, according to datanet.co.kr. Unlike a general chatbot, the model returns results as probabilities, choices and scores rather than generating sentences.
The model is built on Kakao's existing multimodal language model Kanana-2-V-3B. On a single H200 GPU it records average response times under 0.04 seconds, and Kakao states it is 43 times faster than GPT-6 Luna.
Kakao says the model scored highest among the compared models on its Kanana Decision Bench, a set of 15,909 Korean questions, and on the public Decision Index dataset. The comparison set included Microsoft's Decision-1, TypeSafe AI's Zeph, Perplexity's pplx-decider-v1.1-27B and Upstage's Solar Decide.
License and first deployments
The model is released under the Kanana Open License, which permits commercial use. Kakao plans to apply it first to internal projects such as web search and automatic product category classification.
Kakao had already developed Kanana-2-V-3B, the base model behind the new release, before this launch.
Quick answers
What is Kanana-2-V-Decide?
It is Kakao's multimodal decision model, which outputs probabilities, choices and scores instead of generated sentences. It was released as open source on Hugging Face on November 6, 2026.
Can Kanana-2-V-Decide be used commercially?
Yes. It is released under the Kanana Open License, which allows commercial use.
How fast is Kanana-2-V-Decide?
On a single H200 GPU it records average response times under 0.04 seconds, and Kakao states it is 43 times faster than GPT-6 Luna.