Mistral-Nemo-Base-2407: benchmark results
Mistral AI and NVIDIA's Mistral-Nemo-Base-2407, a 12B base model with a 128K context window, trained on multilingual and code data as a drop-in replacement for Mistral 7B. Provider: Mistral. Released 2024-07-18. Access: Open.
Aggregate Rating 1399 ± 29, rank #1228 of 1656 rated models, from 53 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Portuguese LLM - OAB Exams | 56.13 | Accuracy (%) | 89.4 |
| Open Portuguese LLM - ENEM | 72.92 | Accuracy (%) | 84.1 |
| GSM8K | 84.2 | Accuracy (%) | 83.9 |
| Open PL LLM - PolQA Open Book (generative, 5-shot) | 88.43 | Levenshtein Similarity (%) | 83.9 |
| PIQA | 83.5 | Accuracy (%) | 81.7 |
| Open Portuguese LLM - BLUEX | 61.75 | Accuracy (%) | 81 |
| Open PL LLM - 8TAGS (multiple choice, 5-shot) | 77.22 | Accuracy (%) | 75.2 |
| Open PL LLM - PolQA Closed Book (generative, 5-shot) | 66.98 | Levenshtein Similarity (%) | 74.5 |
| Open PL LLM - 8TAGS (generative, 5-shot) | 75.87 | Accuracy (%) | 71.4 |
| Open PL LLM - PoQuAD Open Book (generative, 5-shot) | 67.19 | Levenshtein Similarity (%) | 69.3 |
| Open PL LLM - Belebele (multiple choice, 5-shot) | 85 | Accuracy (%) | 69.1 |
| Open PL LLM - Belebele (generative, 5-shot) | 84.78 | Accuracy (%) | 68.7 |
Interactive version: theaggregate.ai/model?slug=mistral-nemo-base-2407 · How It Works · Data refreshed daily, snapshot 2026-10-20.