Use AI without sending every task to a public service.
Run supported models on your own computer, server, or private Erebrus node, then make them available to approved devices on your trusted network.
Your deployment determines where prompts are processed, what is logged, and which devices can connect.
Choose where models run and who can reach them.
Models on trusted hardware
Download supported models and run them on your own hardware so prompt processing follows the deployment you choose.
Model hub
Browse a curated catalog from edge-friendly 0.5B models up to desktop-class 32B parameter models.
Custom personas
Build system prompts, tune temperature, top-P, max tokens, and stop sequences for repeatable outputs.
LAN node discovery
Your desktop broadcasts _erebrus-ai._tcp over mDNS so phones and other computers find it instantly.
Network model sharing
Use models served by nearby devices, or access private models shared by your workspace.
Wallet sign-in
One Erebrus account across VPN, Firewall, Drop, and AI — authenticate with your existing wallet.
Get Erebrus AI on your device
Available on macOS, Windows, Linux, Android, and iOS. Download a model, pick a persona, and start chatting.
Plan the model and hardware together.
Model licenses and hardware requirements vary. Smaller or heavily quantized models can trade quality for speed and memory efficiency. Local hosting does not automatically make a workflow compliant, accurate, or free from logging; those outcomes depend on the model, configuration, and operating practices.