dolphin-2.9.3-mistral-nemo-12B: benchmark results
Eric Hartford's uncensored Dolphin fine-tune of Mistral Nemo 12B with conversational, coding, and function-calling skills. Provider: Cognitive Computations. Released 2024-07-23. Access: Open.
Aggregate Rating 1514 ± 19, rank #744 of 1661 rated models, from 21 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Ko-LLM Leaderboard | 48.25 | Average (%) | 95.8 |
| Open Korean LLM Leaderboard | 48.25 | Average Score (%) | 95.8 |
| Open Korean LLM - Ko-GPQA-Diamond | 27.78 | Normalized accuracy (%) | 95 |
| Open Korean LLM - Ko-EQ-Bench | 49.47 | EQ-Bench score | 94.9 |
| Open Korean LLM - Ko-Winogrande | 68.51 | Accuracy (%) | 93.3 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 58.38 | Exact match, flexible extract (%) | 92.4 |
| Open Portuguese LLM - FaQuAD NLI | 81.28 | Macro F1 (%) | 91.7 |
| Open Korean LLM - Ko-IFEval | 43.65 | Strict accuracy, prompt/instruction mean (%) | 88 |
| Open Portuguese LLM - ASSIN2 RTE | 93.08 | Macro F1 (%) | 86.1 |
| Open Portuguese LLM - OAB Exams | 52.71 | Accuracy (%) | 82.5 |
| Open Portuguese LLM - BLUEX | 62.45 | Accuracy (%) | 82.4 |
| Open Portuguese LLM - ENEM | 72.08 | Accuracy (%) | 82.1 |
Interactive version: theaggregate.ai/model?slug=dolphin-2-9-3-mistral-nemo-12b · How It Works · Data refreshed daily, snapshot 2026-10-23.