AI21’s Jamba model integrated with Microsoft Azure AI, with 256K context window and 3X throughput
TEL AVIV, Israel, May 21, 2024 /PRNewswire/ — AI21, a leader in building foundation models and AI systems for the enterprise proudly unveils a new collaboration with Microsoft that will bring AI21’s latest cutting-edge Jamba-Instruct model to the Microsoft Azure AI as part of its Model-as-a-Service (MaaS) offering. This collaboration enables Azure customers to be the first to benefit from AI21’s new foundation models benefitting from a novel architecture approach that drives quality, performance, and long context windows on a much smaller and more cost-effective hardware.
Jamba-Instruct, the latest innovation from AI21, is the first production-grade LLM based on the revolutionary hybrid Mamba-Transformer architecture, which AI21 brought to market first. With its unique blend of quality output, high throughput, and a 256K context window, Jamba-Instruct sets a new standard for enterprise-grade AI solutions.
Traditional LLMs use the Transformer architecture, which faces two challenges: high memory usage and slow inference with growing context, limiting practical production use.
Azure customers will soon have access to Jamba-Instruct through Models-as-a-Service without the need to manage underlying infrastructure to host the model, making Azure the first hyperscaler cloud to offer this model to customers. This strategic integration underscores Microsoft’s commitment to providing state-of-the-art AI tools and technologies for enterprises to drive innovation and digital transformation across industries.
Jamba-Instruct is designed to deliver value, quality, and performance for enterprise applications. With its large context window, capable of processing the equivalent of a 400-page novel, Jamba-Instruct excels in tasks such as question answering, document summarization, and building chatbots for sustained and reference-based conversations.
Jamba-Instruct outperforms its competitors in both quality and cost, making it the ideal choice for enterprises seeking advanced AI solutions that prioritize efficiency without compromising performance and reliability.
Jamba-Instruct comes equipped with safety guardrails, chat capabilities, and enhanced command comprehension out of the box. This accelerates the time to production for enterprise applications and reduces TCO, enabling organizations to see a shorter time to value.
“We are thrilled to collaborate with Microsoft to soon introduce Jamba-Instruct to Microsoft Azure customers. This collaboration underscores our shared commitment to empowering enterprises with some of the most innovative AI solutions available,” said Pankaj Dugar, SVP, GM North America of AI21. “Microsoft’s dedication to putting customers first and making cutting-edge technology accessible aligns perfectly with our mission at AI21. With Jamba-Instruct soon available on Azure AI, customers will have access to a transformative tool that will revolutionize how they approach language processing and drive unprecedented levels of efficiency and insight.”
“We’re pleased to collaborate with AI21 to soon introduce the Jamba-Instruct model in our extensive Microsoft Azure AI model catalog offering developers a choice of foundation and open models,” says Eric Boyd, Corporate Vice President, Azure AI Platform at Microsoft. “With its unique architecture and window context length, Jamba-Instruct presents developers with a great opportunity to build at a lower cost while maintaining outstanding quality and performance.”
About AI21
AI21 builds Foundation Models and AI Systems for the enterprise that accelerate the use of GenAI in production. Founded in 2017, AI21 has raised a total of $336 million from NVIDIA, Intel, Google, and others. It was among the first companies to bring generative AI to the masses and, to date, offers consumer applications and enterprise solutions. With the AI21 Platform, businesses can build their own generative AI-driven applications and services. For more information, please visit www.AI21.com.
SOURCE AI21
