One file. One command. Your AI assistant + multi-agent runtime, on your hardware. Connects to free local LLMs (Ollama) or cloud APIs (OpenAI / Anthropic).
Pick your OS — copy, paste, done. The installer takes 30 seconds and creates a Start Menu shortcut.
Windows
PowerShell one-liner. Installs to %LOCALAPPDATA%\\Aether, adds Start Menu + Desktop shortcuts, opens the dashboard automatically.
iwr https://glixylabs.com/install.ps1 | iex
%LOCALAPPDATA%\Aether (no admin needed)localhost:7777 on first runmacOS
Single-line install via curl. Creates ~/.aether and a CLI in ~/.local/bin/aether.
curl -fsSL https://glixylabs.com/install.sh | sh
brew install python)aether CLI to ~/.local/binLinux
Same install script as macOS. Or pip-install directly if you prefer.
curl -fsSL https://glixylabs.com/install.sh | sh
pip install glixy-aether && aether start
aether-runtime/Docker
Run Aether in an isolated container. Mount your data dir to persist memory across upgrades.
docker run -p 7777:7777 -v $HOME/.aether:/data glixylabs/aether
4 GB RAM · 2 GB free disk · any modern CPU. No GPU required (uses Ollama on CPU or cloud APIs).
16 GB RAM · NVIDIA RTX 3060 / Apple Silicon M1 or newer · 30 GB disk for model weights.
NVIDIA A100 / 4090 · 64 GB RAM · 200 GB NVMe. Or use Glixy hosted GPUs (₹2,499/user/mo).
Python 3.10+ (auto-detected). Optional: Ollama for free local LLMs, or any OpenAI-compatible API key.
start.bat (Windows) or ./start.sh (mac/linux). Aether opens at localhost:7777.# Windows · PowerShell PS> iwr https://glixylabs.com/install.ps1 | iex [aether] Python found: Python 3.12.1 [aether] Installing to %LOCALAPPDATA%\Aether [aether] Downloading runtime ... [aether] Installing dependencies ... [aether] ✓ Created Start Menu shortcut [aether] ✓ Created Desktop shortcut # Launch PS> aether start ✦ Glixy Aether v0.4.2 → http://127.0.0.1:7777 → provider: ollama model: llama3 → data: %LOCALAPPDATA%\Aether # Browser opens automatically. # Or visit http://localhost:7777
Yes. The runtime is MIT-licensed. The Pro hosted plan (₹2,499/user/mo) is optional — for teams who want managed GPUs and SSO. Self-hosting is free forever.
No, unless you choose a cloud model. By default Aether runs your conversations through your local Ollama install. Memory and chat history live in a SQLite file on your machine.
No — small models (Llama 3 8B, Phi-3) run on CPU. For 70B-class models you'll want a GPU or use an API key. The runtime auto-detects what's available.
Delete %LOCALAPPDATA%\\Aether on Windows or ~/.aether on mac/linux. That's everything — no registry, no global Python packages.
Re-run the install script. It overwrites the runtime, keeps your data and memory.
Yes. Aether speaks the OpenAI-compatible chat API. Point it at Ollama, vLLM, OpenRouter, Together, Anyscale, your own server — it just works.
Native. Drop an MCP server config in ~/.aether/mcp.json and its tools show up in the dashboard.
One command. Your private AI. Free forever.
iwr https://glixylabs.com/install.ps1 | iex
copy