Red Hat Enterprise Linux AI (RHEL AI)
A bootable Linux image with IBM Granite models and the InstructLab tuning tools already inside. It is the shortest route from bare metal to a working model, and it is where Granite and Red Hat stop being two separate decisions.
RHEL AI is the flat-pack version. Rather than choosing a Linux, choosing a model, choosing a serving engine, and choosing tuning tools as four separate decisions, you boot one image and the four pieces are already fitted together.
What is in the box: a bootable Red Hat Enterprise Linux image, IBM Granite models, and InstructLab. Red Hat built it around the LAB method that came out of IBM Research, which stands for Large-scale Alignment for chatBots and is the technique InstructLab uses to teach a model new material.
Where it fits in a Power project
Treat RHEL AI as the place you learn, and Red Hat AI Inference Server as the place you run. RHEL AI is excellent for getting a model onto your own hardware, feeding it your own documents, and finding out whether the idea holds water at all. It is not the thing IBM names as a Spyre card requirement.
If the pilot works and you then buy a card, the serving decision becomes Red Hat AI Inference Server or Red Hat OpenShift AI. The Granite model you tuned during the pilot carries across, which is the practical benefit of everything in this family being the same open model underneath.
Red Hat's AI portfolio has moved quickly and product names have shifted more than once. Confirm the current packaging and the supported IBM Power configurations with Red Hat before treating this as your production plan.
Related
InstructLab
The open source project that lets the person who knows your business teach a model, without that person being a data scientist. For a shop whose knowledge lives in three people's heads, this is the interesting one.
IBM Granite Models
IBM's own family of open language models, and the cargo an accelerator card is built to carry. You can download them free, run them on your own hardware, and nobody meters the tokens.
Red Hat AI Inference Server
One of the two products IBM names as required before a Spyre card will do anything. It is the driver, the traffic controller, and the reason a 128 GB card can serve more than one person at a time.
Red Hat OpenShift AI
The other product IBM accepts for driving a Spyre card. It is the heavier option: a full container platform for teams running several models, several projects, and more than one server.