Red Hat launches Generative AI Foundation Model Platform for Hybrid Cloud Environment

Open source software giant Red Hat has officially launched its Red Hat Enterprise Linux AI (RHEL AI) platform, designed specifically for generative artificial intelligence (AI) development. The platform, which aims to support generative AI development in hybrid cloud environments, was first announced three months ago at the Red Hat Summit and is now available for use.

RHEL AI integrates IBM's large-scale open source language model, the Granite series, as well as Red Hat's own InstructLab model alignment tool, and is provided in optimized, bootable RHEL image form for deployment on individual servers in hybrid cloud and on-premises environments.

The platform aims to facilitate the development of smaller, more cost-effective generative AI models by providing a more accessible approach. While generative AI technology holds great promise, its high costs often deter adoption. According to Red Hat, the training costs alone for some leading generative AI models can reach up to $200 million, not including data and alignment costs, which are necessary to adapt the models to specific user tasks.

Red Hat believes that the future trend will be towards smaller generative AI models, with the majority adopting an open source approach, similar to the widespread development model of modern software. RHEL AI aims to be the preferred platform for building these low-cost, open source generative AI models. The platform offers a range of freely available Granite large language models (LLMs), which have been pre-compensated and aligned for a variety of common generative AI use cases.

The InstructLab tool helps users fine-tune models using proprietary knowledge and data, without requiring advanced data science skills. With RHEL AI, businesses can train and deploy generative AI at any location in the hybrid cloud, close to their data storage. Additionally, the platform provides a pathway for migration to Red Hat's OpenShift AI platform, enabling the training, tuning, and serving of generative AI models using the same tools and concepts.

RHEL AI is optimized for Dell Technologies' latest PowerEdge servers and offers all the benefits of a Red Hat subscription, including product distribution, 24/7 support, extended model lifecycle support, and model intellectual property compensation, as well as open source legal protection.

As an extension of Red Hat Enterprise Linux, RHEL AI can run in various IT environments, including on-premises servers and cloud infrastructure platforms such as Amazon Web Services (AWS), Google Cloud, IBM Cloud, and Microsoft Azure. Currently, the platform is available as a "bring-your-own-subscription" offering on AWS, IBM Cloud, and on-premises servers, with plans to launch on Google Cloud and Azure later this year, and as a "as-a-service" offering on IBM Cloud next year.