Red Hat Enterprise Linux AI datasheet

July 16, 2026•
Resource type: Datasheet
Print to PDF

Overview

Red Hat® Enterprise Linux® AI is a foundation model platform for consistently running large language models (LLMs) in individual server environments.  It is an optimized, immutable appliance built for inference, taking advantage of Red Hat AI Inference capabilities and image mode for Red Hat Enterprise Linux. By packaging the operating system (OS) and application together, Red Hat Enterprise Linux AI facilitates Day 1 operations to optimize model inference across hybrid cloud environments by maximizing throughput, minimizing latency, and reducing compute costs.

Efficient, cost-effective inference from Day 1

Red Hat Enterprise Linux AI simplifies deployment across hybrid cloud environments by providing a consistent platform to run models where organizations need them—in datacenters, public clouds, or at the edge of the network. Organizations can deploy enterprise open source Granite family models with full Red Hat support and indemnification. 

With Red Hat AI Inference at its core, the platform delivers fast, cost-effective inference by optimizing graphics processing unit (GPU) utilization with vLLM, a high-performance inference runtime that compresses models for increased efficiency, letting users run any AI model on any hardware accelerator across environments. The platform runs on bootable Red Hat Enterprise Linux images via image mode, including PyTorch, required runtime libraries, and hardware accelerator support. This bootable appliance approach can shorten the time from proof of concept to server-based deployments in production.

An expansive catalog offers validated, optimized, open model options for a variety of use cases and requirements. This curated collection accelerates time to value by providing efficient models that are ready for production deployment. Advanced model compression capabilities help reduce hardware requirements and costs through techniques like quantization and speculative decoding, applied to both foundational and custom models.

The platform exposes generative AI (gen AI)-specific telemetry to existing monitoring dashboards, including:

  • Time-to-first-token (TTFT).
  • Key-value (KV)-cache hit rate.
  • Throughput.
  • GPU use. 

This visibility gives organizations the operational transparency they need to meet service-level objectives and control costs.

Red Hat AI Inference, included in Red Hat Enterprise Linux AI, also supports deployment on Red Hat OpenShift® and third-party Kubernetes distributions.

At a glance

Red Hat Enterprise Linux AI lets you run any model on any accelerator and cloud. It includes:

  • Enterprise Granite family LLMs with Red Hat support and indemnification.

  • Validated, optimized open models ready for deployment.

  • Bootable Red Hat Enterprise Linux images optimized for inference.

  • Red Hat AI Inference capabilities for fast, cost-effective model serving.

  • LLM compressor for efficiency while maintaining accuracy.

Related resources

Red Hat AI Hugging Face repository

Why you should care about AI inference

Table 1. Key benefits

Benefit

Description

Efficient, open source gen AI models

  • Access enterprise open source Granite family models, fully indemnified by Red Hat, from the Hugging Face repository.
  • Take advantage of cost- and performance-optimized models aligned to a wide variety of gen AI use cases.
  • Benefit from models released under Apache 2.0 license with transparency on training data sets.

Cost-effective, fast inference

  • Maximize GPU use with vLLM inference runtime to boost throughput and minimize latency.
  • Compress models through techniques like quantization while maintaining response accuracy.

Reduced operational complexity

  • Deploy a simplified, consistent platform across hybrid cloud environments with Day 1 readiness.
  • Manage advanced AI techniques through a user-friendly interface.
  • Accelerate production deployment with a bootable appliance approach via Red Hat Enterprise Linux image mode.

Components

Red Hat Enterprise Linux AI provides an optimized platform that delivers everything needed to run LLMs in production server environments:

  • Inference runtime for hybrid cloud environments. With vLLM at its core, Red Hat AI Inference provides a unified, high-performance platform to run your choice of models across various accelerators. 
  • LLM compressor. Compress both foundational and trained models to reduce compute requirements and related costs while maintaining high model response accuracy. Organizations receive Red Hat support for their model optimization initiatives.
  • Optimized model repository. Access the Red Hat AI repository on Hugging Face. This curated collection of maintained models is ready for inference deployment, including support for both vLLM and NVIDIA NIMs.
  • Bootable model runtime instances. Deploy Granite models packaged as bootable Red Hat Enterprise Linux images via image mode, including PyTorch, required runtime libraries, and hardware accelerator support. This immutable appliance approach helps you deliver consistent, Day 1-ready deployments.
  • Enterprise Granite models. Access open source Granite  models that are fully supported and indemnified by Red Hat, providing cost- and performance-optimized options for diverse gen AI use cases.
  • Enterprise support and lifecycle promise. Starting with a trusted enterprise product distribution, Red Hat provides 24x7 production support and model intellectual property indemnification.

Get started with Red Hat Enterprise Linux AI

Learn how Red Hat Enterprise Linux AI helps organizations expand token economics,, and run their preferred choice of models on any accelerator and in any cloud environment.

Tags:AI inference