Odin-9B: benchmark results
Delta-Vector's creative-writing and roleplay fine-tune of Gemma 2 9B, an earlier checkpoint of the magnum v4 9B training run. Provider: Delta-Vector. Released 2024-09-27. Access: Open.
Aggregate Rating 1513 ± 20, rank #746 of 1661 rated models, from 21 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - Ko-EQ-Bench | 55.55 | EQ-Bench score | 96.6 |
| Open Ko-LLM Leaderboard | 48.53 | Average (%) | 96 |
| Open Korean LLM Leaderboard | 48.53 | Average Score (%) | 96 |
| Open Korean LLM - Ko-IFEval | 53.01 | Strict accuracy, prompt/instruction mean (%) | 95.5 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 58.53 | Exact match, flexible extract (%) | 92.5 |
| Open LLM Leaderboard - MuSR | 46.48 | Score | 89.5 |
| Open LLM Leaderboard - GPQA | 34.14 | Score | 84.9 |
| Open Korean LLM - KorNAT-CKA | 29.21 | Normalized accuracy (%) | 78.8 |
| Open LLM Leaderboard - MMLU-Pro | 40.47 | Score | 74.2 |
| Open LLM Leaderboard - BBH | 54.4 | Score | 70.7 |
| Open Korean LLM - Ko-GPQA-Diamond | 24.75 | Normalized accuracy (%) | 69.5 |
| Open Korean LLM - Ko-Winogrande | 61.48 | Accuracy (%) | 62.4 |
Interactive version: theaggregate.ai/model?slug=odin-9b · How It Works · Data refreshed daily, snapshot 2026-10-23.