Open Korean LLM - KorNAT-CKA: leaderboard
Metric: Normalized accuracy (%). Source: huggingface.co. 1162 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | T3Q-ko-solar-dpo-v7.0 | 50.77 |
| 2 | Myrrh_solar_10.7b_3.0 | 48.57 |
| 3 | lambda-qwen2.5-14B-dpo-test | 46.25 |
| 4 | SOLAR-10.7B-Instruct-T3Q-ko-solar-dpo-ties | 44.39 |
| 5 | SOLAR-10.7B-Instruct-STOCK_SOLAR-ties | 44.12 |
| 6 | llama3.1-korean-v1.1-sft-by-aidx | 42.06 |
| 7 | mistralopithecus-v3-dpo-7B | 40.93 |
| 8 | Dear_My_best_Friends-13B | 40.45 |
| 9 | Qwen2.5-Lumen-14B | 40.1 |
| 10 | COKAL-DPO_test-v2 | 40 |
| 11 | DataVortexS-10.7B-dpo-v1.10 | 39.95 |
| 12 | Qwen2.5-14B-Gutenberg-Instruct-Slerpeno | 39.85 |
| 13 | free-solar-evo-v0.13 | 39.81 |
| 14 | Qwen2.5-14B-Gutenberg-1e-Delta | 39.68 |
| 15 | qwen2.5-reinstruct-alternate-lumen-14B | 39.66 |
Interactive version: theaggregate.ai/benchmark?slug=open-korean-llm-kornat-cka · How It Works · Data refreshed daily, snapshot 2026-09-19.