Llama 3 8B Ko Sft V1: benchmark results
Provider: Meta. Access: Open.
Unified ELO 1495 ± 17, rank #1143 of 2656 rated models, from 10 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM Leaderboard | 41.71 | Average Score (%) | 83.4 |
| Open Ko-LLM Leaderboard | 41.71 | Average (%) | 83.3 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 48.98 | Exact match, flexible extract (%) | 82.5 |
| Open Korean LLM - Ko-IFEval | 37.26 | Strict accuracy, prompt/instruction mean (%) | 77.7 |
| Open Korean LLM - Ko-EQ-Bench | 25.96 | EQ-Bench score | 72.3 |
| Open Korean LLM - KorNAT-CKA | 24.8 | Normalized accuracy (%) | 53.6 |
| Open Korean LLM - KorNAT-Harmless | 62.91 | Normalized accuracy (%) | 51.7 |
| Open Korean LLM - Ko-Winogrande | 60.14 | Accuracy (%) | 51.6 |
| Open Korean LLM - KorNAT-Helpful | 46.33 | Normalized accuracy (%) | 44.8 |
| Open Korean LLM - Ko-GPQA-Diamond | 21.21 | Normalized accuracy (%) | 24 |
Interactive version: theaggregate.ai/model?slug=llama-3-8b-ko-sft-v1 · How It Works · Data refreshed daily, snapshot 2026-09-19.