---
title: "ARX OS"
description: "Release notes for each ARX OS version, newest first"
canonical_url: "https://docs.ardge.com/en/release/arx-os"
---
# ARX OS

> Release notes for each ARX OS version, newest first

ARX OS releases are listed below, newest first.

<u-changelog-versions>
<u-changelog-version date="2026-09-21" title="ARX OS 1.0.1" badge="Latest">
<template v-slot:body="">

ARX OS 1.0.1 refines the initial network setup and introduces the Lab category in App Market.

**Enhancements**

- The setup wizard carries the interface's current subnet mask into the static IP form, so switching to a static address no longer requires the mask to be entered from scratch
- App Market adds the Lab category. Lab hosts ARDGE's experimental AI applications, released on a rolling basis and open to try as soon as they are published

</template>
</u-changelog-version>

<u-changelog-version date="2026-08-28" title="ARX OS 1.0.0">
<template v-slot:body="">

ARX OS 1.0.0 is the first general release of ARX OS. It provides the components required to operate an edge AI platform: model deployment, application management, storage expansion and system administration.

**Initial setup**

A setup wizard runs at first boot and covers the system update, the hostname and administrator account, the time zone, the network and the storage configuration.

**Desktop**

A single sign-in point that presents every application the user has permission for.

- Applications open embedded in the desktop or in a separate browser tab
- The toolbar launches applications and switches between them
- Global search locates system settings and applications directly
- A guided tour runs at first sign-in. **About** shows the device model and the ARX OS version
- Shutdown stops applications and system services in sequence

**AI Model Center**

Deploys and manages inference models on the local GPU.

- **Model Hub** provides open-source models verified by ARDGE, with deployment parameters preconfigured for direct deployment. The Hugging Face tab supports searching for and installing other fine-tuned GGUF or Safetensors models
- **Model Instances** allows inference parameters to be configured, on the vLLM, NVIDIA vLLM and Llama.cpp engines. A VRAM check runs before deployment and blocks instances that exceed the available memory
- **API Key** scopes access to all models or to specified models, with permanent validity or a defined expiry date
- **API Endpoint** is compatible with the OpenAI and Claude API formats, for integration with existing AI workflows

**App Market**

A central source for generative AI applications, ready for use once installation completes.

- **Explore** lists every application published to the market
- **My Applications** manages the installed applications

**Storage Manager**

Manages local storage and connects remote storage to extend the data capacity available to applications.

- **Mount Points** connects remote storage over NFS, so application data can reside remotely without being limited by local capacity
- **Files** browses the contents of a mount point to confirm the system reads the remote data correctly
- **Application Storage Settings** defines where each application reads and writes data

**System Monitor**

- Real-time GPU, CPU, network and disk readings: utilisation, VRAM, temperature, throughput and capacity
- The top 20 processes by resource consumption, with their CPU, memory, GPU and VRAM usage

**Log Center**

Records system and application access and event logs centrally, in a single format.

- Connection sources and system operations are traceable, with system and application records queried from one interface

**Control Panel**

The central location for system administration settings.

- **System Information** shows the device details, live hardware readings, the storage usage breakdown, and the ports used by system services
- **General Settings** and **Network** configure the hostname, the system time, network addressing and interface bonding
- **Users and Roles** manages accounts and roles, with one role per account, the built-in Super Admin and Member roles, and the single Owner identity
- **Security** blocks a source IP when failed sign-ins reach the configured threshold, with allow and deny lists for the exceptions
- **Update** updates the system components and the applications
- **Notification** sends email to administrators when system events occur. Messages follow the system language, and events of the same type within one minute are combined into a single email
- **ARDGE Account** binds the device to an ARDGE cloud account for device, model and application licensing. **Internet Account** stores the Docker Hub credentials so rate limits do not interrupt updates
- **System Reset** erases all data, configuration and installed applications and returns the device to its factory state

</template>
</u-changelog-version>
</u-changelog-versions>


## Sitemap

See the full [sitemap](https://docs.ardge.com/sitemap.md) for all pages.