Reka, an AI research and product company, has selected Oracle Cloud Infrastructure (OCI) to support its mission to advance science and build generative AI models for the benefit of humanity and enterprises. Working in collaboration with Oracle, Reka will also make its multimodal models, supporting over 20 languages and trained on OCI AI Infrastructure powered by NVIDIA GPUs, available in the Oracle Cloud Marketplace.

“We are partnering with Oracle to bring the power of generative AI to more organizations around the world,” said Dani Yogatama, chief executive officer and co-founder, Reka. “OCI’s high-performance AI infrastructure capabilities, dedicated engineering support, and global footprint will continue to enable us to innovate and ship new models quickly and efficiently.”

Founded by a team of highly experienced scientists and engineers from DeepMind, Google Brain, and FAIR, Reka has developed leading multimodal AI models that enable organizations to deploy generative AI applications using inputs from multilingual text, code, images, and videos. These applications can generate concepts for creative tasks, find answers to a range of questions, and derive insights from large datasets. Additionally, Reka’s models enable the development of AI agents that improve as well as enable a wide range of new business use cases through analysis and direct action. By working with Oracle, Reka plans to continue to expand its global reach and advance its focus on enterprise-grade AI products and solutions.

“Reka is pushing performance boundaries by training its multimodal foundation models on OCI Supercluster,” said Greg Pavlik, senior vice president, AI and Data Management Services, Oracle Cloud Infrastructure. “We are looking forward to working with the Reka team to deliver it’s multimodal models, with capabilities that stretch beyond text, to Oracle customers around the world.”

Reka Core and Reka Flash models will be available through Oracle Cloud Marketplace, a centralized repository of enterprise applications offered by Oracle and Oracle partners, via private offers.

Model builders harness the high-performance power of OCI for faster AI deployments

Reka, like many other AI startups and enterprises, is building and training models faster and more reliably with OCI’s purpose-built AI capabilities. For training large language models (LLMs), OCI Supercluster can scale up to 4,096 OCI Compute bare metal instances with 32,768 NVIDIA H100 Tensor Core GPUs that are connected by ultra-low-latency RDMA cluster networking and a choice of HPC storage. OCI Compute virtual machines and OCI’s bare metal NVIDIA GPU instances can power applications for generative AI, computer vision, natural language processing, recommendation systems, and more. Oracle’s dedicated engineering support team works with customers through the entire deployment, from planning to launch, to ensure success.

Additional Resources

• Learn more about OCI AI Infrastructure
• Learn more about accelerating GPU workloads
• Learn more about Oracle Cloud Marketplace
• Learn more about OCI AI ISV Ecosystem