All four tools live in the "private / self-hosted AI" space, but they solve different problems. Pick based on what's already true about your setup.
| Reefy | Ollama | LM Studio | Open WebUI | |
|---|---|---|---|---|
| Shape | Full OS for AI computers | Local LLM runtime | Desktop LLM app | Web UI in front of an LLM runtime |
| Bring own hardware | Boot from USB on any x86_64 | Install on your existing OS | Install on Mac/Windows/Linux desktop | Install on your existing server |
| OS setup required | None (Reefy is the OS) | You install + maintain Linux/macOS/Windows | You install + maintain your desktop OS | You install + maintain Linux + Docker |
| GPU drivers | NVIDIA baked into firmware | You install separately | You install separately | You install separately |
| App catalog | OpenClaw, Hermes, Ollama, vLLM, SGLang, qr-access, ... | LLM models only | LLM models only | Chat UI only |
| Multi-instance isolation | Per-app containers | Single process | Single process | Single process |
| Remote access | Built-in Cloudflared tunnels | You configure | You configure | You configure |
| A/B firmware + watchdog | Yes | N/A (app, not OS) | N/A | N/A |
| Encrypted cloud backups | Built-in (BorgBase) | You configure | You configure | You configure |
| BYO LLM subscription | Built-in (reefy-llm-proxy: Codex, xAI) | No | No | No |
| Cost | Free, open source | Free, open source | Free for personal | Free, open source |
When to pick Reefy
- You want to repurpose a PC (old laptop, mini-PC, NUC, used server) as a private AI box without learning Linux.
- You want to run more than one app: an LLM, an agent, a vision pipeline, a door-access system - all isolated and managed from one dashboard.
- You want BYO LLM subscription so you don't pay per-token API bills.
- The device will live in a closet/attic and needs to survive bad updates and brief power loss without an on-site reboot.
When to pick Ollama or LM Studio instead
- You already have a tuned Linux/Mac and just need a local LLM runtime; the OS+orchestration parts of Reefy would be redundant.
- You want a desktop chat experience on your laptop (LM Studio).
When to pick Open WebUI instead
- You already have an LLM runtime (Ollama, vLLM, etc.) and just want a chat UI on top.
You can also run Ollama or Open WebUI inside Reefy via its app catalog - they're not mutually exclusive. Reefy is the host, they're the apps.