llama3.1-korean-v1.3-sft-by-aidx: benchmark results
Provider: Meta. Access: Open.
Unified ELO 1537 ± 17, rank #908 of 2656 rated models, from 10 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - KorNAT-CKA | 33.72 | Normalized accuracy (%) | 93 |
| Open Korean LLM Leaderboard | 45.92 | Average Score (%) | 92.3 |
| Open Ko-LLM Leaderboard | 45.92 | Average (%) | 92.2 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 56.18 | Exact match, flexible extract (%) | 90.2 |
| Open Korean LLM - Ko-EQ-Bench | 36.28 | EQ-Bench score | 86.4 |
| Open Korean LLM - Ko-IFEval | 42.21 | Strict accuracy, prompt/instruction mean (%) | 85.9 |
| Open Korean LLM - KorNAT-Helpful | 47.73 | Normalized accuracy (%) | 59.4 |
| Open Korean LLM - Ko-Winogrande | 60.85 | Accuracy (%) | 58 |
| Open Korean LLM - KorNAT-Harmless | 63.09 | Normalized accuracy (%) | 55.4 |
| Open Korean LLM - Ko-GPQA-Diamond | 21.72 | Normalized accuracy (%) | 29.4 |
Interactive version: theaggregate.ai/model?slug=llama3-1-korean-v1-3-sft-by-aidx · How It Works · Data refreshed daily, snapshot 2026-09-19.