koOpenChat-sft: benchmark results
Provider: maywell. Access: Open.
Aggregate Rating 1370 ± 27, rank #7178 of 10916 rated models, from 15 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - KorNAT-Harmless | 63.8 | Normalized accuracy (%) | 67.1 |
| Open Korean LLM - Ko-Winogrande | 62.04 | Accuracy (%) | 65.8 |
| Open Korean LLM - KorNAT-Helpful | 48.23 | Normalized accuracy (%) | 65.8 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 29.42 | Exact match, flexible extract (%) | 63.7 |
| Open Korean LLM - Ko-EQ-Bench | 15.17 | EQ-Bench score | 60.8 |
| Open Korean LLM Leaderboard | 36.73 | Average Score (%) | 58.4 |
| Open LLM Leaderboard v1 - MMLU | 61.32 | Accuracy (%) (5-shot) | 51.5 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 51.24 | MC2 (%) (0-shot) | 51.2 |
| Open LLM Leaderboard v1 - WinoGrande | 76.4 | Accuracy (%) (5-shot) | 48 |
| Open LLM Leaderboard v1 - ARC Challenge | 59.81 | Normalized accuracy (%) (25-shot) | 45 |
| Open LLM Leaderboard v1 - GSM8K | 24.18 | Accuracy (%) (5-shot) | 43.4 |
| Open Korean LLM - Ko-IFEval | 26.69 | Strict accuracy, prompt/instruction mean (%) | 39.8 |
Interactive version: theaggregate.ai/model?slug=koopenchat-sft · How It Works · Data refreshed daily, snapshot 2026-11-08.