magnum-v3-27B-kto: benchmark results
Anthracite's KTO-tuned magnum v3 fine-tune of Gemma 2 27B, part of the series aiming to replicate Claude 3 Sonnet/Opus prose quality. Provider: anthracite-org. Released 2024-09-06. Access: Open.
Aggregate Rating 1542 ± 19, rank #641 of 1661 rated models, from 21 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM Leaderboard | 51.06 | Average Score (%) | 97.8 |
| Open Ko-LLM Leaderboard | 51.06 | Average (%) | 97.7 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 68.46 | Exact match, flexible extract (%) | 97.1 |
| Open Korean LLM - Ko-EQ-Bench | 51.34 | EQ-Bench score | 95.5 |
| Open Korean LLM - Ko-GPQA-Diamond | 27.78 | Normalized accuracy (%) | 95 |
| Open Korean LLM - KorNAT-CKA | 34.84 | Normalized accuracy (%) | 95 |
| Open Korean LLM - Ko-IFEval | 48.5 | Strict accuracy, prompt/instruction mean (%) | 92.6 |
| Open Korean LLM - Ko-Winogrande | 67.4 | Accuracy (%) | 91 |
| Open Portuguese LLM - OAB Exams | 56.36 | Accuracy (%) | 89.7 |
| Open LLM Leaderboard - GPQA | 35.57 | Score | 89.6 |
| Open Portuguese LLM - ENEM | 74.88 | Accuracy (%) | 87.7 |
| Open Portuguese LLM - BLUEX | 64.67 | Accuracy (%) | 85.3 |
Interactive version: theaggregate.ai/model?slug=magnum-v3-27b-kto · How It Works · Data refreshed daily, snapshot 2026-10-23.