Red Hat Enterprise Linux 10 is not just a version bump. It represents a fundamental shift in how Red Hat thinks about enterprise Linux β image-based deployments by default, AI workloads as first-class citizens, and a development experience that finally feels modern.
Image Mode by Default
The biggest change in RHEL 10 is the shift to image-based deployments. Instead of managing packages on live systems, you build immutable OS images and deploy them atomically:
# Build a RHEL 10 image
sudo bootc image build --target qcow2 rhel10-base.oci
# Deploy to a system
sudo bootc switch registry.redhat.io/rhel10/rhel-bootc:latestThis approach:
- Eliminates configuration drift between systems
- Makes rollbacks instant (boot the previous image)
- Simplifies compliance (image is the audit artifact)
- Integrates naturally with container and Kubernetes workflows
Traditional package management (dnf) still works, but image mode is the recommended path for new deployments.
AI-Ready Infrastructure
RHEL 10 includes the InstructLab toolkit and vLLM serving runtime, making it the first enterprise Linux distribution designed for AI workloads out of the box:
# Install InstructLab on RHEL 10
sudo dnf install instructlab
# Initialize and start training
ilab model init
ilab data generate
ilab model trainFor GPU workloads, RHEL 10 supports NVIDIA drivers through the official NVIDIA CUDA repository and includes optimized libraries for AI model serving.
Kernel and Performance
RHEL 10 ships with kernel 6.12 LTS, bringing:
- io_uring improvements for high-performance async I/O
- BPF enhancements for observability and networking
- Memory management optimizations for large-memory GPU systems
- Rust kernel modules support (experimental)
- Energy-aware scheduling for power efficiency
Developer Experience
The development toolchain has been modernized:
- GCC 14, LLVM 18, Rust 1.80+
- Python 3.12 as the system Python
- Node.js 22 LTS
- Go 1.22+
- Podman 5.x with improved Docker compatibility
Security Improvements
- SELinux policy improvements for container workloads
- FIPS 140-3 validated cryptographic modules
- Confidential computing support (AMD SEV-SNP, Intel TDX)
- Image signing and verification built into the bootc workflow
Migration from RHEL 9
# In-place upgrade using Leapp
sudo dnf install leapp-upgrade
sudo leapp preupgrade
sudo leapp upgradeThe Leapp tool handles the upgrade path, but for production systems, Red Hat recommends the image-based approach: build a RHEL 10 image, test it, deploy it.
Automating with Ansible
Ansible fully supports RHEL 10. The rhel-system-roles collection has been updated for RHEL 10 compatibility:
---
- name: Configure RHEL 10 systems
hosts: rhel10
roles:
- rhel-system-roles.selinux
- rhel-system-roles.firewall
- rhel-system-roles.crypto_policies
tasks:
- name: Enable AI toolkit
ansible.builtin.dnf:
name: instructlab
state: presentFinal Thoughts
RHEL 10 is Red Hatβs most opinionated release in years. Image mode by default is a bold move that aligns enterprise Linux with how modern infrastructure actually works. If you are running Kubernetes, GPU workloads, or any infrastructure-as-code workflow, RHEL 10βs image-based approach will feel natural. The AI integration makes it particularly relevant for teams deploying models on RHEL AI.