Get DIO — v0.1.0-alpha
Download DIO.
One binary, three platforms. Free. Open source. Works offline out of the box.
Pick your platform
Windows
dio-v0.1.0-alpha-windows-x64.exe
23 MB
Windows (one-click)
install-dio.bat
— launches automatically
Downloads, verifies SHA256, and launches DIO.
Linux x64
dio-v0.1.0-alpha-linux-x64
32 MB
Linux (one-click)
install-dio.sh
— launches automatically
curl -sSL https://avadios.com/downloads/install-dio.sh | bash · downloads, verifies SHA256, launches DIO.
macOS
coming soon
— in progress
Native .dmg in progress. Email hello@avadios.com to be notified the moment it ships.
Every binary is signed and ships with a SHA256 sidecar. macOS .dmg and Linux AppImage / .deb are next on the roadmap — drop a note to hello@avadios.com if you want to be notified.
One-line install
Windows · PowerShell
iwr https://avadios.com/downloads/install-dio.ps1 -UseBasicParsing | iex
Linux · bash
curl -sSL https://avadios.com/downloads/install-dio.sh | bash
Or run from a USB drive
DIO is portable. Drop the binary on any USB stick (FAT32 / exFAT) and run it directly. No install, no admin rights, no traces. The bundle includes the binary, a starter local model, and an INSTALL.sh / INSTALL.bat for permanent setup.
First five minutes
- 1
Open DIO.
It detects your hardware and picks a default local model.
- 2
Set your contribution.
Slide CPU, RAM, storage, bandwidth to whatever you are comfortable sharing. Or none — DIO works fully solo.
- 3
Pick your brains.
Choose local-only, or paste API keys for Claude, OpenAI, Gemini, Grok, or DeepSeek.
- 4
Say hi.
"Hey DIO" — voice is on by default. Or type into Chat.
- 5
Optional: connect to AVADIOS.
Toggle network mode in Contribute. Earn AUTOM from cycle one.
System requirements
| Minimum | Recommended | |
|---|---|---|
| OS | Windows 10 / macOS 12 / Linux (kernel 5.4+) | Latest stable release |
| CPU | x86-64, 4 cores | x86-64 or ARM64, 8+ cores |
| RAM | 8 GB | 16 GB+ |
| Disk | 2 GB free | 20 GB+ for larger local models |
| GPU | None required | NVIDIA + CUDA 11/12 for ~10× faster local inference |