AI21 Brings Hybrid SSM-Transformer Model Jamba to Snowflake Cortex AI

Partnership will democratize access to AI21's high-performing model for all Snowflake customers, seamlessly integrating with the Snowflake AI Data Cloud

SAN FRANCISCO, June 4, 2024 /PRNewswire/ -- AI21, a leader in AI systems for enterprises, today announced at Snowflake Data Cloud Summit 2024, that it is bringing its newest model, Jamba-Instruct, the instruction-tuned version of Jamba, built from a novel hybrid SSM-Transformer architecture to the Snowflake AI Data Cloud. This partnership will enable all Snowflake customers to access AI21's Jamba-Instruct for immediate and secure enterprise use.

Jamba-Instruct will soon be accessible in Snowflake Cortex AI, Snowflake's comprehensive managed service offering access to top-tier large language models and vector search capabilities. AI21's model will integrate seamlessly with the Snowflake AI Data Cloud, making it simple for customers with varying levels of expertise to interact with AI21's models, whether they are specialized AI engineers or SQL developers. Cortex AI streamlines the deployment of generative AI projects through its sophisticated management of GPU infrastructure, significantly expediting the development timeline for customers using an AI21 model.

"We are excited to be working with Cortex AI to bring our models to more customers around the world who are eager to innovate with AI technologies," said Ori Goshen, AI21's CEO and Co-founder. "With Cortex AI's serverless experience, data security, and governance, we know our customers will be in good hands."

This partnership follows AI21's launch of Jamba, the world's first production-grade Mamba-style model that is setting a new standard for efficiency and performance. With the largest context window in its size class of 256K, Jamba outperforms other state-of-the-art competitive models across a wide range of benchmarks including efficiency, quality, and scalability. With this large context window, capable of processing the equivalent of a 400-page novel, Jamba-Instruct, the enterprise-ready version of Jamba, excels in tasks such as question answering, document summarization, and building chatbots for sustained and reference-based conversations. Deploying Jamba-Instruct on an accessible platform like Cortex AI will enable high-powered, valuable long context use cases for any Snowflake customer.

"With AI21's groundbreaking Jamba-Instruct offered in Cortex AI our customers will now be able to seamlessly connect their data to build transformative GenAI applications with AI21's powerful models," said Baris Gultekin, Head of AI at Snowflake. "We are excited to further our partnership with AI21, driven by our shared dedication to simplifying and ensuring the reliability of Gen AI solutions for enterprises. Looking ahead, we anticipate incorporating more of AI21's model offerings into Snowflake Cortex AI."

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