AI Controller
AI Controller
Enterprise on-premise GPU and AI model management with the Qubrid AI Controller
An enterprise software solution for your on-premise servers. The Qubrid AI Controller gives IT administrators a single pane of glass into GPU infrastructure with the ability to:
- Control GPU resource and AI model allocation for enterprise users
- Monitor usage across nodes and workloads
- Update drivers and maintain bare-metal AI environments
- Install AI packages, deploy models, and manage Docker workloads centrally
End users get their own easy-to-use interface with access to the latest models, programming environments, and containers - all running on your organization's on-prem servers.
Evaluation: You can try this software on your existing servers. The evaluation is limited to 16 GPUs and a 60-day trial period.
Download the controller
Download the installation package and follow the Installation guide to set up the primary node on Ubuntu.
qubrid_packages.tar.bz2
DownloadThe package contains .deb installers for Ubuntu 22.04 LTS and Ubuntu 24.04 LTS (primary and worker node packages). You can also request access from the AI Controller tab on the Qubrid AI Platform.
Installation guide
Step-by-step setup: system requirements, install commands, first login, and adding worker nodes.
Who uses what
| Role | What you get |
|---|---|
| Administrators | Dashboard for nodes, hardware monitoring, GPU history, user management, model deployment, compute and image management |
| Enterprise users | Personal console with assigned GPUs, models, Docker images, and programming environments on shared on-prem infrastructure |
Documentation in this section
Installation
Minimum requirements, supported GPUs, download, install, first access, and worker nodes.
Dashboard
Monitor nodes, system and network info, hardware metrics, GPU history, and AI package installation.
Model Studio
Deploy LLMs on your premises, download Hugging Face and NIM models, and run GPU-accelerated inference.
Compute Management
Add and delete nodes, download container images, and assign Docker images to users.
User Management
Add users, set passwords and expiry, allocate GPU/CPU resources, and manage permissions.
Troubleshooting
Reset admin and user passwords from the primary node SSH terminal.
Before you install:
- Ubuntu 22.04 or 24.04 LTS on your host(s)
- At least 16 GB RAM and 10 CPU cores on the primary node
- Python 3.10+, NVIDIA GPU with supported drivers
- Static IP addresses on all nodes and open ports for controller and user workloads (see Installation)