LG K-EXAONE 2.0 Card - KMMLU-Pro: leaderboard
KMMLU-Pro professional-level Korean knowledge questions, run by LG AI Research. LG AI Research K-EXAONE 2.0 model card main table (July 2026), which its technical report (arXiv 2608.04505, Table 6) gives as reasoning mode for all five models; accuracy in percent, higher is better.
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Around February 2027. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek V4 Pro (Max) | 80.5 | #112 (DeepSeek V4 Pro) |
| 2 | Qwen 3.5 397B A17B (Reasoning) | 77.4 | #154 (Qwen 3.5 397B A17B) |
| 3 | GLM-5.1 (Thinking) | 75.8 | #109 (GLM-5.1) |
| 4 | K-EXAONE (Reasoning) | 67.3 | #387 (K-EXAONE) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=lg-k-exaone-2-0-card-kmmlu-pro · How It Works · Data refreshed daily, snapshot 2026-11-05.