janus-7B: benchmark results
KAIST's Janus-7B, a Mistral 7B v0.2 fine-tune supervised-fine-tuned on all 196K entries of a system-message preference dataset for diverse, personalized alignment. Provider: KAIST. Released 2024-04-04. Access: Open.
Aggregate Rating 1376 ± 30, rank #1316 of 1656 rated models, from 19 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard - MuSR | 44.01 | Score | 78.2 |
| Open Portuguese LLM - FaQuAD NLI | 72.39 | Macro F1 (%) | 65.5 |
| Open Portuguese LLM - OAB Exams | 42.69 | Accuracy (%) | 52.9 |
| Open Portuguese LLM - ENEM | 62.21 | Accuracy (%) | 51.1 |
| Open Portuguese LLM - BLUEX | 50.21 | Accuracy (%) | 47 |
| Open Chinese LLM - TruthfulQA MC | 53.25 | Accuracy (%) | 46.7 |
| Open Portuguese LLM - ASSIN2 RTE | 88.3 | Macro F1 (%) | 44.2 |
| Open LLM Leaderboard - BBH | 46.94 | Score | 38.9 |
| Open LLM Leaderboard - IFEval | 37.75 | Score | 36.8 |
| Open LLM Leaderboard - MMLU-Pro | 28.74 | Score | 32.6 |
| Open LLM Leaderboard - GPQA | 27.27 | Score | 27.6 |
| Open Chinese LLM - GSM8K | 31.46 | Accuracy (%) | 25 |
Interactive version: theaggregate.ai/model?slug=janus-7b · How It Works · Data refreshed daily, snapshot 2026-10-20.