LogicKor — leaderboard

Korean multi-domain reasoning benchmark leaderboard covering reasoning, math, writing, coding, comprehension, grammar, single-turn, and multi-turn tasks.

Metric: Score (0-10). Source: lk.instruct.kr. Status: saturated. 253 models tracked.

Top models

#ModelScore
1GPT-4o (2024-05-13)9.41
2Claude 3.5 Sonnet (20240620)9.3
3Gemini 1.5 Pro (001)9.29
4GPT-4 Preview (0125)9.29
5GPT-4 Preview (1106)9.17
6GPT-4 Turbo9.09
7Mistral Large 2 (Jul)9.03
8Gemini 1.5 Flash (001)8.98
9Claude 3 Opus (20240229)8.96
10Llama-VARCO-8B-Instruct8.83
11Llama 3.1 405B Instruct FP88.79
12GPT-4 (0613)8.67
13ko-gemma-2-9B-it8.67
14Qwen 2 72B Instruct8.61
15Claude 3 Sonnet (20240229)8.38

Interactive version: theaggregate.ai/benchmark?slug=logickor · How the rankings work · Data refreshed daily, snapshot 2026-07-22.