Mistral-7B-Base-SFT-KTO: benchmark results
Princeton NLP's Mistral-7B-Base-SFT-KTO, a Mistral 7B base model with SFT and KTO alignment, released alongside Princeton's SimPO preference-optimization paper. Provider: Princeton NLP. Released 2024-05-17. Access: Open.
Aggregate Rating 1402 ± 19, rank #1226 of 1661 rated models, from 21 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - Ko-IFEval | 37.07 | Strict accuracy, prompt/instruction mean (%) | 77.6 |
| Open LLM Leaderboard - MuSR | 43.68 | Score | 75.6 |
| Open Korean LLM - Ko-GPQA-Diamond | 24.75 | Normalized accuracy (%) | 69.5 |
| Open Korean LLM - Ko-EQ-Bench | 23.86 | EQ-Bench score | 69.2 |
| Open Ko-LLM Leaderboard | 37.73 | Average (%) | 63.7 |
| Open Korean LLM Leaderboard | 37.73 | Average Score (%) | 63.7 |
| Open Portuguese LLM - ASSIN2 RTE | 90.4 | Macro F1 (%) | 57.2 |
| Open LLM Leaderboard - IFEval | 47.85 | Score | 55.6 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 22.37 | Exact match, flexible extract (%) | 54.9 |
| Open Portuguese LLM - FaQuAD NLI | 66.15 | Macro F1 (%) | 51.8 |
| Open Korean LLM - KorNAT-Harmless | 62.68 | Normalized accuracy (%) | 48.9 |
| Open Portuguese LLM - OAB Exams | 41.41 | Accuracy (%) | 47.1 |
Interactive version: theaggregate.ai/model?slug=mistral-7b-base-sft-kto · How It Works · Data refreshed daily, snapshot 2026-10-23.