RiosGateway tunnels AI requests from any app to the Ollama running on your machine — or straight to Ollama Cloud. No port forwarding, no exposed ports, one API key.
POST /users/<you>/<node>/v1/chat/completions
A public server receives every request, checks your API key, applies rate limits, and logs usage.
Your PC connects outbound over WebSocket. Nothing listens on your router — no ports opened, no firewalls touched.
Requests run on your local Ollama and stream back token by token. Cloud mode skips the node and calls Ollama Cloud directly.
Only allowed origins pass CORS. Prompts and responses are never stored — only model names, timing, and token counts.
Register on this dashboard and create a client to generate your API key.
Run it on the same machine as your local Ollama, or skip it entirely if you use Cloud mode.
Paste your username, slug, and key into the app. The tunnel connects on launch.
Use your endpoint URL and API key in any OpenAI-compatible app, and pick a model.
The gateway, dashboard, and clients are fully open source. Don't trust the binary — read it.
The client is an unsigned Electron executable. Every release is scanned publicly.
🛡️ 0/72 detections — v1.3.0