Kakao on the 28th released four lightweight Kanana 2 models via the global AI platform Hugging Face, consisting of base and instruct models with 1.3 billion and 3 billion parameters, all covered by the Kanana Open License for commercial use. The small language model (SLM) can run on smartphones and personal computers, aiming to secure technological competitiveness in On-device AI and expand the foundation for Korea's development ecosystem. Kakao equipped the models with a Korean-specialized tokenizer, increasing processing efficiency by more than 30% compared with before. It also applied a sliding window attention technology that, while processing conversations up to 32,000 tokens long, can cut memory usage by up to 72.7%. The company said it secured performance competitive with similar-sized models such as Qwen and Gemma in key evaluations including Korean and English, math, coding, and tool use. Kakao is using the lightweight AI models for KakaoTalk chat and call summaries, Kanana in KakaoTalk, and the AI National Secretary.