Command is Cohere’s family of generative language models, aimed squarely at enterprise use rather than consumer chat. The line came into focus in 2024 with Command R and the larger Command R+, both designed around the workloads Cohere treats as central to business deployments: retrieval-augmented generation (RAG), tool use, low-latency high-throughput serving, and strong performance across roughly ten key languages. Command R offers a 128,000-token context length, useful for grounding answers in long documents.
Cohere iterates the family with dated releases. The August 2024 update (command-r-08-2024) delivered roughly 50 percent higher throughput and 20 percent lower latency than the prior version while halving the hardware needed to serve it, and Command R+ in the same refresh was tuned for more complex, nuanced tasks with the same long context. Cohere has continued the line with newer flagship models, some released with open weights, so the specific members named here reflect Cohere’s announcements as of the verification date.
Why business readers should care: Command is one of the few model families engineered first for enterprise RAG and tool use rather than general chat, with a deliberate focus on grounding, multilingual coverage, and the cost of serving at production scale.