Llumen

---

Why Llumen?

Most self-hosted interfaces are built for servers, not devices. They're powerful but demand heavy resources and hours of configuration.

Llumen carves out a different space: privacy without the complexity. You get the features you actually need, optimized for modest hardware: Raspberry Pi/old laptops/minimal VPS while keeping many features of commercial products.

	Privacy	Power	Setup
Commercial (ChatGPT)	❌ Cloud-only	✅ High	✅ Zero-config
Typical Self-Host (Open WebUI)	✅ Local	✅ High	❌ Config hell
llumen	✅ Local	⚖️ Just enough	✅ Zero-config

Features

Feature	What You Get
Speed	Sub-second cold starts, real-time token streaming
Chat Modes	Normal, Web Search, & Deep Research with autonomous agents
Rich Media	PDF uploads, LaTeX rendering, image generation
Universal API	Any OpenAI-compatible provider (OpenRouter, local models, etc.)
Minimal Footprint	~17MB binary, <128MB RAM usage

video.mp4

Quickstart

Default Login: admin / P@88w0rd

Docker (30-second setup)

docker run -it --rm \
  -e API_KEY="<YOUR_OPENROUTER_KEY>" \
  -p 80:80 \
  -v "$(pwd)/data:/data" \
  ghcr.io/pinkfuwa/llumen:latest

That's it. No config files. No Python dependencies.

Want bleeding edge? Use ghcr.io/pinkfuwa/llumen:nightly

See ./docs/sample for docker-compose examples.

Native Binaries

Download from Releases for Windows/Linux, including arm build.

Configuration (Optional)

Variable	Description	Default
API_KEY	OpenRouter/provider key	required
API_BASE	Custom endpoint	https://openrouter.ai/api
DATA_PATH	Storage directory	.
BIND_ADDR	Network Socket	0.0.0.0:80

Documentation

https://pinkfuwa.github.io/llumen/

Built with ❤️ by pinkfuwa. Keep it simple, keep it fast.