gemma-ko-v01: benchmark results
Provider: Google. Access: API.
Unified ELO 1355 ± 17, rank #2200 of 2656 rated models, from 10 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - KorNAT-Harmless | 65.45 | Normalized accuracy (%) | 80.4 |
| Open Korean LLM - Ko-IFEval | 28.2 | Strict accuracy, prompt/instruction mean (%) | 46.7 |
| Open Korean LLM - KorNAT-CKA | 22.67 | Normalized accuracy (%) | 40.8 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 5.69 | Exact match, flexible extract (%) | 25 |
| Open Korean LLM - Ko-GPQA-Diamond | 21.21 | Normalized accuracy (%) | 24 |
| Open Korean LLM Leaderboard | 30.78 | Average Score (%) | 21.5 |
| Open Ko-LLM Leaderboard | 30.78 | Average (%) | 21.4 |
| Open Korean LLM - KorNAT-Helpful | 43.03 | Normalized accuracy (%) | 21.1 |
| Open Korean LLM - Ko-EQ-Bench | -0.64 | EQ-Bench score | 4.5 |
| Open Korean LLM - Ko-Winogrande | 49.17 | Accuracy (%) | 1.2 |
Interactive version: theaggregate.ai/model?slug=gemma-ko-v01 · How It Works · Data refreshed daily, snapshot 2026-09-19.