Mistral NeMo Instruct

A state-of-the-art 12B multilingual model with a 128k context window, designed for global applications and strong in multiple languages.

Benchmark results

Benchmark Score Tags Source
CommonSenseQA 70.4% self-reported llm-stats link →
HellaSwag 83.5% self-reported llm-stats link →
MMLU 68.0% self-reported llm-stats link →
Natural Questions 31.2% self-reported llm-stats link →
OpenBookQA 60.6% self-reported llm-stats link →
TriviaQA 73.8% self-reported llm-stats link →
TruthfulQA 50.3% self-reported llm-stats link →
Winogrande 76.8% self-reported llm-stats link →