gemma-ko-7B-instruct-v0.50: benchmark results
Provider: Google. Access: API.
Aggregate Rating 1324 ± 29, rank #8258 of 10916 rated models, from 14 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - KorNAT-CKA | 39.55 | Normalized accuracy (%) | 98.5 |
| Open Korean LLM - Ko-Winogrande | 65.9 | Accuracy (%) | 86.8 |
| Open Korean LLM - Ko-GPQA-Diamond | 24.75 | Normalized accuracy (%) | 69.5 |
| Open Korean LLM - Ko-IFEval | 32.58 | Strict accuracy, prompt/instruction mean (%) | 62.9 |
| Open Korean LLM Leaderboard | 36.85 | Average Score (%) | 59.1 |
| Open Korean LLM - Ko-EQ-Bench | 6.91 | EQ-Bench score | 52.9 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 13.65 | Exact match, flexible extract (%) | 41.5 |
| Open LLM Leaderboard v1 - GSM8K | 12.28 | Accuracy (%) (5-shot) | 34 |
| Open LLM Leaderboard v1 - WinoGrande | 72.61 | Accuracy (%) (5-shot) | 29.5 |
| Open LLM Leaderboard v1 - MMLU | 49.58 | Accuracy (%) (5-shot) | 28.9 |
| Open LLM Leaderboard v1 - HellaSwag | 75.21 | Normalized accuracy (%) (10-shot) | 26.1 |
| Open Korean LLM - KorNAT-Harmless | 61.23 | Normalized accuracy (%) | 25.3 |
Interactive version: theaggregate.ai/model?slug=gemma-ko-7b-instruct-v0-50 · How It Works · Data refreshed daily, snapshot 2026-11-08.