AI21’s Jamba 1.6: The Best Open Model for Private Enterprise Deployment

AI21 Labs has introduced Jamba 1.6, an open model family designed for enterprise deployment that claims to outperform models like Mistral Large 2 and Llama 3.3 70B. Featuring a hybrid SSM-Transformer architecture and a 256K context window, the model excels at Retrieval-Augmented Generation (RAG) and long-context tasks. Jamba 1.6 offers flexible deployment options, including on-premise or in a private cloud, addressing enterprise data security and privacy concerns. Customer use cases demonstrate improvements in data classification and grounded question answering, and a new Batch API is available for efficiently handling large volumes of asynchronous requests.