Microsoft Partners with Cohere; Command R+ Joins Azure AI Model Hub
Microsoft recently announced that it will collaborate with Cohere to integrate its latest developed large-scale language model, Command R+, into its cloud platform. This integration will allow Cohere's cutting-edge model to be included in the expanding model library of Azure AI Studio, providing developers with an efficient tool for building intelligent applications.
Command R+ is the newest open-weight model in the Cohere family, with enhanced retrieval-augmented generation (RAG) and optimized tool usage capabilities. It can assist enterprises in building customized and precise AI solutions based on internal data and documents. The model's output includes clear references, effectively reducing the risk of hallucination and providing additional context from the source material.
Aidan Gomez, CEO and co-founder of Cohere, said, "We are at a critical period in the development of AI, and collaborations between industry leaders like Cohere and Microsoft will truly shape the future of enterprise AI. Command R+ represents not only a significant technological leap but also our steadfast commitment to ethical, scalable, and impactful AI solutions."
The integration of this model with tools in Azure AI Studio, such as Azure AI Content Safety, Azure AI Search, and prompt flow, will further enhance the responsibility and efficiency of AI practices. Its key advantages include enhanced security and compliance, integration with content safety, simplified LLM process evaluation, and convenient deployment and inference.
Microsoft provides the Cohere model through the "Model as a Service" (MaaS) approach, simplifying the usage process and enabling enterprises to fully leverage the powerful capabilities of Command R+ without worrying about the complexity of underlying infrastructure. This pay-as-you-go model offers flexibility and scalability for businesses of different scales.
Including Command R+ in the Azure AI model library is part of Microsoft's ongoing efforts to provide comprehensive, cutting-edge, and open model choices. Currently, the model library already has over 1600 foundational models from industry-leading companies such as Databricks, Deci AI, Hugging Face, Meta, Microsoft Research, Mistral AI, NVIDIA, OpenAI, and Stability AI.