Kanana-2-v-decide: Kakao's open decision model answers in 0.04 seconds
Kakao released Kanana-2-v-decide, a 3.67B multimodal decision model that reads text and images and returns typed judgments (choice, yes/no probability, or score) instead of generated text, averaging under 0.04 seconds on a single H200 GPU; open under the Kanana Open License for commercial use.
Why it matters
It replaces a frontier model call for every routine agent judgment: typed probability/choice/score outputs at 0.04s, which Kakao claims is 43x faster than GPT-6 Luna on the same decisions, with top scores on the public Decision Index benchmark.
What you could build with it
A customer-support team could put this 3B decision model in front of its agent fleet to classify incoming tickets and pick the right tool or escalation path in under 0.04 seconds, instead of paying a frontier model to reason over every routine judgment.
Does it hold up?
Too early to judge: released today, weights are on Hugging Face under a commercial-use license with Transformers code in the model card, but no independent deployments exist yet.
Built with Kanana-2-v-decide
- Kakao Unveils Agent AI Model Specialized in Judgment, 43 Times Faster Than GPT-6edaily · Korean press report on the Nov 6 release: 0.04s average response on one H200, Kanana Open License, top results on Kanana Decision Bench and Decision Index.
- Kakao Unveils AI Model for Text and Image-Based Decision-Makingthelec · Release coverage: the model classifies customer inquiries and picks agent tools, built on Kanana-2-v-3b, with probability, choice and score outputs.
Learn more
First spotted on hf: source.
More AI models
VoiceStudio: open-source, fully-local ElevenLabs alternative for voice cloning, dubbing, and audiobooksdebpalash/VoiceStudio is a desktop app that runs voice cloning, voice design, video dubbing, dictation, transcription…models · JEV 0.75Soup: fine-tune an 8B LLM on a 4 GB laptop GPU from one YAML fileAn open-source CLI that fine-tunes LLMs from a single YAML file, with layer streaming that trains an 8B model on a 4…models · JEV 0.73Tencent Hunyuan Hy3: open 295B MoE with 21B active under Apache 2.0Tencent open-sourced Hy3, a 295B-parameter mixture-of-experts model with only 21B parameters active per token, under…models · JEV 0.73NVIDIA Isaac Lab 3.0 GA + Isaac OS open runtime + Agile ONE SN humanoidMajor Isaac platform expansion at GTC Berlin Oct 21: Isaac Lab 3.0 is generally available as the open-source robot…models · JEV 0.73NVIDIA Nemotron 3.5 Retriever collection: Embed 8B/1B, Rerank 8B, Parse 631MNVIDIA's Oct 21 collection drop for agentic retrieval: Nemotron-3.5-Embed in 8B BF16, 1B BF16 and 1B NVFP4…models · JEV 0.72Kimi K3.1 Pro: Moonshot's 2.8T open-weight flagship officially launchesMoonshot AI officially launched Kimi K3.1 Pro on Oct 23, a mixture-of-experts model with 2.8 trillion total / 104…models · JEV 0.72
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.