Llama-3-Ko-8B-Instruct-AOG: benchmark results
Provider: Meta. Access: Open.
Unified ELO 1441 ± 17, rank #1585 of 2656 rated models, from 10 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - KorNAT-CKA | 30.73 | Normalized accuracy (%) | 84.8 |
| Open Korean LLM - Ko-Winogrande | 62.98 | Accuracy (%) | 71.6 |
| Open Korean LLM - KorNAT-Helpful | 48.7 | Normalized accuracy (%) | 71.5 |
| Open Korean LLM - KorNAT-Harmless | 63.47 | Normalized accuracy (%) | 63.2 |
| Open Ko-LLM Leaderboard | 36.78 | Average (%) | 59 |
| Open Korean LLM Leaderboard | 36.78 | Average Score (%) | 59 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 24.87 | Exact match, flexible extract (%) | 58.2 |
| Open Korean LLM - Ko-EQ-Bench | 4.04 | EQ-Bench score | 50.3 |
| Open Korean LLM - Ko-IFEval | 25.01 | Strict accuracy, prompt/instruction mean (%) | 32.4 |
| Open Korean LLM - Ko-GPQA-Diamond | 21.72 | Normalized accuracy (%) | 29.4 |
Interactive version: theaggregate.ai/model?slug=llama-3-ko-8b-instruct-aog · How It Works · Data refreshed daily, snapshot 2026-09-19.