Kanana-Nano-2.1B
Kakao's smallest bilingual Korean-English language model, fully open source.
About
Kanana-Nano-2.1B is the nano tier of Kakao's Kanana bilingual language model series, released open source in February 2025 alongside a technical report. It targets Korean and English with competitive performance for its size class, using compute-efficient techniques including staged pre-training, depth up-scaling, and distillation.
The model ships in base, instruct, embedding, function-call, and RAG variants to support research on Korean language models. Kakao notes that neither pre-training nor post-training data includes Kakao user data.
The Kanana series spans 2.1B to 32.5B parameters; the 2.1B nano models are the only fully open tier, while larger Flag and Essence tiers remain internal. It is a practical pick for Korean-language experimentation on modest hardware.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 104; Trending score 68/100 from researcher assessment.