Command R: Definition and Examples

Command R is a family of large language models developed by Cohere, specifically optimized for professional use cases such as retrieval-augmented generation (RAG) and tool use.

Full definition

Command R is a series of language models created by the Canadian company Cohere, launched in March 2024. Unlike general-purpose models like GPT-4 or Claude, Command R was designed from the start to excel in enterprise applications, with a particular focus on retrieval-augmented generation (RAG), tool use, and processing long documents.

The Command R family includes several variants: Command R, the base model offering excellent performance-to-cost ratio, and Command R+, a more powerful version for complex tasks. These models stand out for their ability to precisely cite their sources when generating answers, which significantly reduces hallucinations and enhances result reliability in professional contexts.

A major feature of Command R is its extended context window of 128,000 tokens, allowing it to process very long documents. The model also supports over 10 languages, making it a relevant choice for international companies. Its architecture is optimized for efficient deployment, with competitive inference costs compared to similarly sized models.

Command R is accessible via Cohere's API, as well as through third-party platforms like Amazon Bedrock, Microsoft Azure, and Hugging Face. Cohere has also released Command R model weights under a permissive license, allowing companies to deploy it on their own infrastructure for use cases requiring data sovereignty.

Etymology

The name "Command" refers to Cohere's lineage of models dedicated to text generation (as opposed to their "Embed" model for embeddings). The suffix "R" denotes this new generation optimized for RAG (Retrieval-Augmented Generation), marking a departure from previous Command versions.

Concrete examples

Corporate document search with RAG

Using the provided documents, answer the following question by precisely citing your sources: What are the termination conditions of the supplier contract signed in 2024?

Tool use to automate workflows

You have access to the following tools: search_crm, send_email, calendar. A client requested a meeting next week. Find their contact details in the CRM and suggest three available time slots.

Multilingual synthesis of long documents

Here is an 80-page annual report in English. Generate a 500-word executive summary in French, identifying the 5 key points and main risks mentioned in the document.

Practical usage

In prompt engineering, Command R is particularly suitable when building RAG systems that require reliable citations. Leverage its native tool use capability to orchestrate complex workflows without needing external frameworks. For best results, structure your source documents with clear metadata, as the model excels at identifying and citing relevant passages.

Related concepts

RAG (Retrieval-Augmented Generation)Tool UseContext WindowEmbeddings

FAQ

What is the difference between Command R and Command R+?

Command R is the base model, optimized for cost/performance on RAG and tool use tasks. Command R+ is a larger, more powerful version designed for complex tasks requiring advanced reasoning. In practice, Command R is sufficient for most RAG use cases, while Command R+ is recommended for multi-document analysis and multi-step reasoning.

Is Command R open source?

Cohere has published Command R's weights on Hugging Face under a CC-BY-NC license allowing research and evaluation. For commercial use, a specific license from Cohere is required. The model can be self-hosted, which interests companies with data sovereignty constraints.

How does Command R compare to GPT-4 or Claude for RAG?

Command R was specifically trained for RAG with source citation, making it a natural choice for this use case. GPT-4 and Claude are more general-purpose models offering better performance in general reasoning and creative generation. The choice depends on the use case: for a production RAG system needing source traceability, Command R is very competitive; for varied tasks requiring advanced reasoning, general-purpose models often remain preferable.

How to use this prompt

Copy the prompt with the button above.
Paste it into ChatGPT, Claude or your favorite AI assistant.
Replace the bracketed variables with your details, then refine the result.

About Prompt Guide

Prompt Guide is a free library of 2500+ ready-to-use prompts for ChatGPT, Claude and other AIs, with guides to learn prompting and tools to build and optimize your own prompts.

Prompt library Learn prompting Prompt builder Prompt optimizer

More definitions

Completion: Definition and Examples

Response generated by a language model (LLM) from a given prompt. Completion is the text produced by the AI to complete, answer, or extend the user's input.

Computer Use: Definition and Examples

Ability of an AI model to directly interact with a computer by controlling the mouse, keyboard, and screen, just as a human user would.

Confusion Matrix: Definition and Examples

Learn to read a confusion matrix: true positives, false negatives, accuracy, precision and recall explained with concrete examples in machine learning.

Constitutional AI: Definition and Examples

AI alignment method developed by Anthropic, where a model is trained to self-correct by following a set of written principles (a 'constitution')

Context Window: Definition and Examples

The context window refers to the maximum amount of text a language model can process at one time, encompassing both the user input and the generated response.

Continual Learning: Definition and Examples

Continual Learning refers to the ability of an AI model to learn new tasks or data sequentially, without forgetting previously acquired knowledge.

Get new prompts every week

Join our newsletter.