Llama 1 3B Test02: benchmark results
Provider: Meta. Access: Open.
Unified ELO 1386 ± 17, rank #2012 of 2656 rated models, from 10 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - Ko-Winogrande | 62.43 | Accuracy (%) | 68.5 |
| Open Korean LLM - KorNAT-Harmless | 63.11 | Normalized accuracy (%) | 56.2 |
| Open Korean LLM - KorNAT-Helpful | 47.05 | Normalized accuracy (%) | 52.5 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 18.73 | Exact match, flexible extract (%) | 50 |
| Open Korean LLM - Ko-IFEval | 25.32 | Strict accuracy, prompt/instruction mean (%) | 34.5 |
| Open Korean LLM Leaderboard | 32.43 | Average Score (%) | 33.5 |
| Open Ko-LLM Leaderboard | 32.43 | Average (%) | 33.2 |
| Open Korean LLM - KorNAT-CKA | 21.17 | Normalized accuracy (%) | 33.1 |
| Open Korean LLM - Ko-EQ-Bench | 0 | EQ-Bench score | 24.5 |
| Open Korean LLM - Ko-GPQA-Diamond | 16.67 | Normalized accuracy (%) | 0.4 |
Interactive version: theaggregate.ai/model?slug=llama-1-3b-test02 · How It Works · Data refreshed daily, snapshot 2026-09-19.