AI Study Online
🌐

Command R+

Cohere · 2024-04

Cohere's enterprise RAG-optimized model designed for scalable retrieval-augmented generation.

Visit Website

Quick Facts

Parameters

Undisclosed (estimated ~100B)

Context Window

128K tokens

Modalities

text

Open Source

No

Pricing

API from $2.50/1M input tokens

Released

2024-04

Developer

Cohere

About

Command R+ is Cohere's flagship enterprise model optimized for retrieval-augmented generation (RAG) workflows. It features a 128K token context window and is specifically designed to excel at grounding responses in external knowledge sources, making it ideal for enterprise search, document analysis, and knowledge management. Command R+ supports multilingual capabilities across 10+ languages, with strong performance in tool use, citation generation, and structured output. Its architecture is optimized for scalable deployment in enterprise environments with robust security and compliance features.

Strengths

  • +Best-in-class RAG performance for enterprise use
  • +Automatic citation generation for grounded responses
  • +Strong multilingual retrieval across 10+ languages
  • +Enterprise-grade security and compliance features

Weaknesses

  • Not available as open-weight model
  • Less capable for creative and general conversation
  • Requires RAG infrastructure for best results

Best For

Enterprise search and knowledge management

Document Q&A with citation grounding

Customer support with retrieval augmentation

Compliance-regulated AI deployments

Pricing

API (Cohere Platform)

From $2.50/1M input tokens

  • RAG-optimized
  • 128K context
  • Citations
  • Tool use
  • Multilingual

Enterprise

Custom pricing

  • Dedicated deployment
  • SLA
  • Compliance
  • Custom fine-tuning

Benchmarks

BenchmarkCommand R+Competitor
RAG (HHEM)92.3%GPT-4: 88.1%

Technical Specs

Parameters

Undisclosed (estimated ~100B)

Context Window

128K tokens

Modalities

text

Languages

EnglishFrenchGermanSpanishItalian+5

Open Source

No

Developer

Cohere

Released: 2024-04

Share this article

Related Models