
Bits With Brains
Curated AI News for Decision-Makers
What Every Senior Decision-Maker Needs to Know About AI and its Impact
Command R+: Cohere's Not-So-Subtle Jab at GPT-4's Enterprise Dominance
4/12/24
Editorial team at Bits with Brains
Cohere Unveils Command R+, a Powerful Enterprise-Grade LLM to Rival GPT-4.

Cohere, a major frontier AI lab, has launched Command R+, its most advanced and scalable large language model (LLM) purpose-built for real-world enterprise applications. Command R+ joins Cohere's R-series of models focused on balancing high efficiency with strong accuracy, enabling businesses to move beyond proof-of-concept and into production with AI.
Command R+ has an impressive array of features tailored for enterprise needs. With a generous 128k-token context window (about 96,000 words), the model excels at advanced retrieval augmented generation (RAG), providing in-line citations to reduce inaccuracies and hallucinations.
Retrieval-Augmented Generation (RAG) is a technique that allows AI language models to retrieve relevant information from external sources and incorporate it into their generated responses, resulting in more accurate, up-to-date, and contextually relevant output. This RAG capability helps enterprises quickly surface the most relevant information from their proprietary data to support tasks across HR, sales, marketing, and customer support.
In addition, Command R+ offers broad multilingual support in 10 key business languages: English, French, Spanish, Italian, German, Portuguese, Japanese, Korean, Arabic, and Chinese. This enables global organizations to seamlessly deploy AI solutions that serve diverse teams and markets.
While comprehensive multilingual benchmarks are still emerging, early indications suggest Command R+ is highly competitive with other top models. In head-to-head comparisons, Command R+ demonstrates performance on par with or better than significantly larger and more expensive models like GPT-4 and Mistral Large.
From a cost perspective, Command R+ also offers compelling value. On Azure, accessing the model costs 50-75% less than GPT-4 per output while generating responses up to 5 times faster.
While not an open-source model, Command R+'s model weights are freely available for research purposes, with commercial licensing available upon request for those seeking on-premises deployment.
The combination of performance and accessibility makes Command R+ a strong contender for enterprises looking to operationalize AI across their organizations. The model's advanced tool functionality, accessible via API, empowers businesses to automate complex workflows by dynamically interacting with external systems like CRMs, databases, and search engines. Multi-step tool use even allows Command R+ to combine multiple tools over several steps and self-correct when encountering errors.
Cohere highlights several enterprise use cases well-suited for Command R+, such as knowledge management assistants that leverage proprietary data, multilingual chatbots for global customer support, and intelligent agents that automate data entry and retrieval across siloed systems. As more businesses seek to transition from experimentation to production AI, Command R+'s balance of accuracy and efficiency at scale should prove attractive.
The release of Command R+ reflects the intensifying competition among tech giants and startups alike to capture enterprise AI market share. Microsoft has already announced Command R+'s availability on the Azure AI platform, expanding its growing portfolio of LLMs optimized for business use. Meanwhile, Cohere seems committed to a cloud-agnostic approach, with upcoming availability on other major cloud platforms as well.
As adoption of enterprise-grade LLMs like Command R+ accelerates, the technology's transformative impact will likely be felt across both private and public sectors. Knowledge workers will increasingly interact with AI co-pilots to boost productivity. Organizations will automate routine cognitive tasks and decision-making. Personalized customer experiences will become the norm, powered by multilingual AI that deeply understands each business' unique domain.
Sources:
[1] https://gigazine.net/gsc_news/en/20240408-command-r-plus-cohere-llm/
[2] https://www.youtube.com/watch?v=6dgXALb_5Ag
[4] https://www.youtube.com/watch?v=zD7NhWrxEy8
[5] https://venturebeat.com/ai/coheres-new-ai-command-r-is-available-on-sagemaker-and-azure/
[6] https://anakin.ai/blog/command-r-coheres-gpt-4/
[7] https://pureai.com/Articles/2024/04/09/Cohere-Command-R-Plus.aspx
[8] https://clarifai.com/cohere/generate/models/command-r-plus
[9] https://aws.amazon.com/marketplace/pp/prodview-jnw3ilzmy7tcg?applicationId=AWSMPContessa&sr=0-2
[11] https://txt.cohere.com/command-r-plus-microsoft-azure/
[12] https://venturebeat.com/ai/coheres-command-r-now-available-on-huggingchat/
[13] https://www.promptfoo.dev/docs/guides/cohere-command-r-benchmark/
[15] https://venturebeat.com/ai/cohere-launches-command-r-a-powerful-llm-optimized-for-enterprise-ai/
[16] https://huggingface.co/CohereForAI/c4ai-command-r-plus
[18] https://the-decoder.com/coheres-new-llm-command-r-is-optimized-for-chatting-with-your-data-rag/
[19] https://www.reddit.com/r/LocalLLaMA/comments/1bvniaz/command_r_cohere_for_ai_104b/
Sources