Yi-Ko-6B: benchmark results
Provider: beomi. Released 2023-12-01. Access: Open.
Aggregate Rating 1326 ± 29, rank #8236 of 10916 rated models, from 15 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - Ko-Winogrande | 60.77 | Accuracy (%) | 57.3 |
| Open Korean LLM - KorNAT-CKA | 22.49 | Normalized accuracy (%) | 39.8 |
| Open LLM Leaderboard v1 - MMLU | 55.72 | Accuracy (%) (5-shot) | 37.1 |
| Open LLM Leaderboard v1 - GSM8K | 12.51 | Accuracy (%) (5-shot) | 34.3 |
| Open Korean LLM - Ko-IFEval | 25.14 | Strict accuracy, prompt/instruction mean (%) | 33.2 |
| Open LLM Leaderboard v1 - WinoGrande | 72.93 | Accuracy (%) (5-shot) | 30.5 |
| Open Korean LLM Leaderboard | 31.63 | Average Score (%) | 28.4 |
| Open LLM Leaderboard v1 - HellaSwag | 74.48 | Normalized accuracy (%) (10-shot) | 24.7 |
| Open Korean LLM - Ko-EQ-Bench | 0 | EQ-Bench score | 24.5 |
| Open LLM Leaderboard v1 - ARC Challenge | 48.89 | Normalized accuracy (%) (25-shot) | 23.6 |
| Open Korean LLM - KorNAT-Harmless | 61.12 | Normalized accuracy (%) | 23.4 |
| Open Korean LLM - KorNAT-Helpful | 43.33 | Normalized accuracy (%) | 23.1 |
Interactive version: theaggregate.ai/model?slug=yi-ko-6b · How It Works · Data refreshed daily, snapshot 2026-11-08.