# Multilingual RAG optimization

Command R7B is an open-weight 7B parameter research version model, optimized for diverse scenarios such as reasoning, summarization, Q&A, and coding, supporting 23 languages.

Command R+ is a 104B parameter open-weight research-grade model developed by Cohere Lab, supporting multilingual and complex task automation.

Command-R is a high-performance 35 billion parameter generative model research version, optimized for scenarios such as reasoning, summarization, and question answering, supporting generation capabilities in 10 languages with exceptional Retrieval-Augmented Generation (RAG) performance.