Red Hat Summit Red Hat SummitSupportConsoleDevelopersStart a trialContact

Chapter 1. Red Hat Enterprise Linux AI 1.5 release notes

RHEL AI provides organizations with a process to develop enterprise applications on open source Large Language Models (LLMs).

1.1. About this release
Copy link

Red Hat Enterprise Linux AI version 1.5 includes various features for Large Language Model (LLM) fine-tuning on the Red Hat and IBM produced Granite model. A customized model using the RHEL AI workflow consisted of the following:

  • Install and launch a RHEL 9.4 instance with the InstructLab tooling.
  • Host information in a Git repository and interact with a Git-based taxonomy of the knowledge you want a model to learn.
  • Run the end-to-end workflow of synthetic data generation (SDG), multi-phase training, and benchmark evaluation.
  • Serve and chat with the newly fine-tuned LLM.

1.2. Features and Enhancements
Copy link

Red Hat Enterprise Linux AI version 1.5 includes various features for Large Language Model (LLM) fine-tuning.

1.2.1. Suppored accelerators
Copy link

1.2.1.1. NVIDIA H200 accelerators
Copy link

You can now use NVIDIA H200 accelerators on RHEL AI version 1.5 for inference serving and running the full end-to-end workflow. When initlizing your RHEL AI environment, select a H200 profile that matches the number of accelerators in your machine. For more information on the supported hardware on RHEL AI, see Red Hat Enterprise Linux AI hardware requirements.

You can now use NVIDIA H200 accelerators on RHEL AI version 1.5 for inference serving in Technology Preview. RHEL AI does not include a system profile for the Grace Hopper accelerators by default. To use the GH200 accelerators, initialize your RHEL AI environment with the h200_x1 profile and add the max_startup_attempts: 1200 parameter to the config.yaml file. 

$ ilab config edit
Copy to Clipboard Toggle word wrap 
serve:
vllm:
    gpus: 1
    llm_family: ''
    max_startup_attempts: 1200
    vllm_args: ["--tensor-parallel-size", "1"]
Copy to Clipboard Toggle word wrap

1.2.1.3. AMD MI300X accelerators
Copy link

AMD MI300X accelerators are now generally available for inference serving and running the full end-to-end workflow. For more information on the supported hardware on RHEL AI, see Red Hat Enterprise Linux AI hardware requirements.

1.2.2. Installing
Copy link

Red Hat Enterprise Linux AI is installable as a bootable image. This image contains various tooling for interacting with RHEL AI. The image includes: Red Hat Enterprise Linux 9.4, Python version 3.11 and InstructLab tools for model fine-tuning. For more information about installing Red Hat Enterprise Linux AI, see Installation overview and the "Installation feature tracker"

1.2.3. Building your RHEL AI environment
Copy link

After installing Red Hat Enterprise Linux AI, you can set up your RHEL AI environment with the InstructLab tools.

1.2.3.1. Initializing InstructLab
Copy link

You can initialize and set up your RHEL AI environment by running the ilab config init command. This command creates the necessary configurations for interacting with RHEL AI and fine-tuning models. It also creates proper directories for your data files. For more information about initializing InstructLab, see the Initialize InstructLab documentation.

1.2.3.2. Downloading Large Language Models
Copy link

You can download various Large Language Models (LLMs) provided by Red Hat to your RHEL AI machine or instance. You can download these models from a Red Hat registry after creating and logging in to your Red Hat registry account. For more information about the supported RHEL AI LLMs, see the Downloading models documentation and the "Large Language Models (LLMs) technology preview status".

1.2.3.2.1. Version 2 of the 3.1 Granite models
Copy link

RHEL AI version 1.5 now supports the granite-3.1-8b-starter-v2 student model and granite-3.1-8b-lab-v2 inference model. For more information about models, see the Downloading Large Lanuage Models documentation.

1.2.3.3. Serving and chatting with models
Copy link

Red Hat Enterprise Linux AI version 1.5 allows you to run a vLLM inference server on various LLMs. The vLLM tool is a memory-efficient inference and serving engine library for LLMs that is included in the RHEL AI image. For more information about serving and chatting with models, see Serving and chatting with the models documentation.

1.2.4. Creating skills and knowledge YAML files
Copy link

On Red Hat Enterprise Linux AI, you can customize your taxonomy tree using custom YAML files so a model can learn domain-specific information. You host your knowledge data in a Git repository and fine-tune a model with that data. For detailed documentation on how to create a knowledge markdown and YAML file, see Customizing your taxonomy tree.

1.2.5. Generating a custom LLM using RHEL AI
Copy link

You can use Red Hat Enterprise Linux AI to customize a granite starter LLM with your domain specific skills and knowledge. RHEL AI includes the LAB enhanced method of Synthetic Data Generation (SDG) and multi-phase training.

1.2.5.1. Synthetic Data Generation (SDG)
Copy link

Red Hat Enterprise Linux AI includes the LAB enhanced method of synthetic data generation (SDG). You can use the qna.yaml files with your own knowledge data to create hundreds of artifical datasets in the SDG process. For more information about running the SDG process, see Generating a new dataset with Synthetic data generation (SDG).

RHEL AI version 1.5 now supports using llama-3.3-70b-Instruct as a teacher model when running Synthetic Data Generation (SDG) as a Technology Preview. For more information, see the Using the llama-3.3-70B-Instruct model as a teacher model (Technology Preview) documentation.

1.2.5.2. Training a model with your data
Copy link

Red Hat Enterprise Linux AI includes the LAB enhanced method of multi-phase training: A fine-tuning strategy where datasets are trained and evaluated in multiple phases to create the best possible model. For more details on multi-phase training, see Training your data on the model.

1.2.5.3. Benchmark evaluation
Copy link

Red Hat Enterprise Linux AI includes the ability to run benchmark evaluations on the newly trained models. On your trained model, you can evaluate how well the model knows the knowledge or skills you added with the MMLU_BRANCH or MT_BENCH_BRANCH benchmark. For more details on benchmark evaluation, see Evaluating your new model.

1.2.6. Red Hat cross-product features
Copy link

You can now run RHEL AI workloads in playbooks using the Ansible Automation Platform hub. This includes two Ansible collections:

infra.ai
A content collection that can provision RHEL AI environments on various cloud providers infrastructure, including AWS, GCP, and Azure. This collection simplifies the deployment of your AI workloads across various cloud providers.
redhat.ai
A content collection designed to manage workloads in RHEL AI. You can use the Ansible playbook options to quickly create deployments in RHEL AI, this allows for more efficient model training and serving processes.

If you are an existing Ansible Automation Platform customer, these collections are included in your current subscription.

1.3. Red Hat Enterprise Linux AI feature tracker
Copy link

Some features in this release are currently in Technology Preview. These experimental features are not intended for production use. For more information about the support scope of Red Hat Technology Preview features, see Technology Preview Features Support Scope.

In the following tables, features are marked with the following statuses:

  • Not Available
  • Technology Preview
  • General Availability
  • Deprecated
  • Removed

1.3.1. Installation feature tracker
Copy link

Expand
Table 1.1. Installation features
Feature1.11.21.31.41.5

Installing on bare metal

Generally available

Generally available

Generally available

Generally available

Generally available

Installing on AWS

Generally available

Generally available

Generally available

Generally available

Generally available

Installing on IBM Cloud

Generally available

Generally available

Generally available

Generally available

Generally available

Installing on GCP

Not available

Technology preview

Generally available

Generally available

Generally available

Installing on Azure

Not available

Generally available

Generally available

Generally available

Generally available

1.3.2. Platform support feature tracker
Copy link

Expand
Table 1.2. End-to-end InstructLab workflow
Feature1.11.21.31.41.5

Bare metal

Generally available

Generally available

Generally available

Generally available

Generally available

AWS

Generally available

Generally available

Generally available

Generally available

Generally available

IBM Cloud

Not available

Generally available

Generally available

Generally available

Generally available

Google Cloud Platform

Not available

Technology preview

Generally available

Generally available

Generally available

Azure

Not available

Generally available

Generally available

Generally available

Generally available

Expand
Table 1.3. Inference serving LLMs
Feature1.11.21.31.41.5

Bare metal

Generally available

Generally available

Generally available

Generally available

Generally available

AWS

Generally available

Generally available

Generally available

Generally available

Generally available

IBM Cloud

Generally available

Generally available

Generally available

Generally available

Generally available

Google Cloud Platform (GCP)

Not available

Technology preview

Generally available

Generally available

Generally available

Azure

Not available

Generally available

Generally available

Generally available

Generally available

Expand
Table 1.4. Cloud Marketplace support
Feature1.11.21.31.41.5

AWS

Not available

Not available

Generally available

Generally available

Generally available

Azure

Not available

Not available

Generally available

Generally available

Generally available

1.4. Large Language Models tracker
Copy link

Expand
Table 1.5. LLM support on hardware vendors

Feature

NVIDIA

AMD

Intel

granite-3.1-8b-starter-v2.1

Generally Available

Generally Available

Not Available

granite-3.1-8b-lab-v2.1

Generally Available

Generally Available

Not Available

granite-3.1-8b-starter-v2

Not Available

Not Available

Technology Preview

granite-3.1-8b-lab-v2

Not Available

Not Available

Technology Preview

granite-8b-code-instruct

Technology preview

Technology Preview

Technology Preview

granite-8b-code-base

Technology Preview

Technology Preview

Technology Preview

mixtral-8x7B-instruct-v0-1

Generally Available

Generally Available

Technology Preview

llama-3.3-70b-Instruct

Technology Preview

Technology Preview

Not Available

prometheus-8x7b-v2.0

Generally Available

Generally Available

Not Available

1.5. Known Issues
Copy link

Running MMLU evaluation

On RHEL AI version 1.5, you need to use the --skip-server flag when running MMLU.

Incorrect auto-detection on some NVIDIA A100 systems

RHEL AI sometimes auto-detects the incorrect system profile on machines with A100 accelerators.

You can select the correct profile by re-initializing and passing the correct system profile. 

$ ilab config init --profile <path-to-system-profile>
Copy to Clipboard Toggle word wrap

Fabric manager does not always starts with NVIDIA accelerators

After installing Red Hat Enterprise Linux AI on NVIDIA systems, you may see the following error when serving or training a model. 

INFO 2024-11-26 22:18:04,244 instructlab.model.serve_backend:56: Using model '/var/home/cloud-user/.local/share/instructlab/phased/phase2/checkpoints/hf_format/samples_29117' with -1 gpu-lay
ers and 4096 max context size.
INFO 2024-11-26 22:18:04,244 instructlab.model.serve_backend:88: '--gpus' flag used alongside '--tensor-parallel-size' in the vllm_args section of the config file. Using value of the --gpus

File "/opt/app-root/lib64/python3.11/site-packages/vllm/entrypoints/openai/api_server.py", line 105, in build_async_engine_client
    async with build_async_engine_client_from_engine_args(
  File "/usr/lib64/python3.11/contextlib.py", line 210, in __aenter__
    return await anext(self.gen)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/opt/app-root/lib64/python3.11/site-packages/vllm/entrypoints/openai/api_server.py", line 192, in build_async_engine_client_from_engine_args
    raise RuntimeError(
RuntimeError: Engine process failed to start
Copy to Clipboard Toggle word wrap

To resolve this issue, you need the run the following commands: 

$ sudo systemctl stop nvidia-persistenced.service
$ sudo systemctl start nvidia-fabricmanager.service
$ sudo systemctl start nvidia-persistenced.service
Copy to Clipboard Toggle word wrap

UI AMD technology preview installations

Red Hat Enterprise Linux AI version 1.5 currently does not support graphical based installation with the technology previewed AMD ISOs. Ensure that the text parameter in your kickstart file is configured for non-interactive installs. You can also pass inst.text in your shell during interactive installation to avoid an install time crash.

SDG can fail on 4xL40s

For SDG to run on 4xL40s, you need to run SDG with the --num-cpus flag and set to the value of 4. 

$ ilab data generate --num-cpus 4
Copy to Clipboard Toggle word wrap

MMLU and MMLU_BRANCH on the granite-8b-starter-v1 model

When evaluating a model built from the granite-8b-starter-v1 LLM, there might an error where vLLM does not start when running the MMLU and MMLU_BRANCH benchmarks.

If vLLM does not start, add the following parameter to the serve section of your config.yaml file: 

serve:
  vllm:
    vllm_args: [--dtype bfloat16]
Copy to Clipboard Toggle word wrap

Kdump over nfs

Red Hat Enterprise Linux AI version 1.5 does not support kdump over nfs without configuration. To use this feature, run the following commands: 

mkdir -p /var/lib/kdump/dracut.conf.d
echo "dracutmodules=''" > /var/lib/kdump/dracut.conf.d/99-kdump.conf
echo "omit_dracutmodules=''" >> /var/lib/kdump/dracut.conf.d/99-kdump.conf
echo "dracut_args --confdir /var/lib/kdump/dracut.conf.d --install /usr/lib/passwd --install /usr/lib/group" >> /etc/kdump.conf
systemctl restart kdump
Copy to Clipboard Toggle word wrap

1.6. Asynchronous z-stream updates
Copy link

Security, bug fix, and enhancement updates for RHEL AI 1.5 are released as asynchronous z-stream updates.

This section will continue to be updated over time to provide notes on enhancements and bug fixes for future asynchronous z-stream releases of RHEL AI 1.5. Versioned asynchronous releases, for example with the form RHEL AI 1.5.z, will be detailed in subsections.

Issued: 11 June 2025

Red Hat Enterprise Linux AI release 1.5.1 is now available. This release includes bug fixes and product enhancements.

1.6.1.1. Features
Copy link

RHEL AI 1.5.1 and further 1.5.z releases support Intel Gaudi 3 accelerators for inference serving models. You can download the Red Hat Enterprise Linux AI image on the Download Red Hat Enterprise Linux AI page and deploy RHEL AI on a machine with Gaudi3 accelerators.

1.6.1.2. Upgrade
Copy link

To update your RHEL AI system to the most recent z-stream version, you must be logged in to the Red Hat registry and run the following command: 

$ sudo bootc upgrade --apply
Copy to Clipboard Toggle word wrap

Issued: 24 June 2025

Red Hat Enterprise Linux AI release 1.5.2 is now available. This release includes bug fixes and product enhancements.

1.6.2.1. Upgrade
Copy link

To update your RHEL AI system to the most recent z-stream version, you must be logged in to the Red Hat registry and run the following command: 

$ sudo bootc upgrade --apply
Copy to Clipboard Toggle word wrap

Issued: 02 September 2025

Red Hat Enterprise Linux AI release 1.5.3 is now available. This release includes bug fixes and product enhancements.

1.6.3.1. Bug Fix
Copy link

Previously, InstructLab was incorrectly detecting the chat template when serving the granite-3.1-8b-lab-v2.2 model on 8xL4 NVIDIA accelerators. This causes serving to fail when attempting to serve the granite-3.1-8b-lab-v2.2 model. The RHEL AI version 1.5.3 now uses the correct chat template and results in successful vLLM deployments on 8xL4 NVIDIA accelerators

1.6.3.2. Upgrade
Copy link

To update your RHEL AI system to the most recent z-stream version, you must be logged in to the Red Hat registry and run the following command: 

$ sudo bootc upgrade --apply
Copy to Clipboard Toggle word wrap
Back to top
Red Hat logoGithubredditYoutubeTwitter

Learn

Try, buy, & sell

Communities

About Red Hat Documentation

We help Red Hat users innovate and achieve their goals with our products and services with content they can trust. Explore our recent updates.

Making open source more inclusive

Red Hat is committed to replacing problematic language in our code, documentation, and web properties. For more details, see the Red Hat Blog.

About Red Hat

We deliver hardened solutions that make it easier for enterprises to work across platforms and environments, from the core datacenter to the network edge.

Theme

© 2025 Red Hat