Skip to content

Roadmap

  • Point one URL at several Ollama machines. Chat goes to the host that already has that model.
  • Use Ollama’s API, OpenAI /v1, or Anthropic /v1/messages (text).
  • Lock the proxy with API keys, or leave it open on a trusted network.
  • List models as the union of every host’s ollama list.
  • Install with curl -fsSL https://orouta.dev/install.sh | sh (needs a GitHub release for the binary).
  • A small web page: which host, which model, errors, how busy
  • ollama pull through orouta onto a machine you choose
  • See when a host is down instead of a silent miss
  • Change the TOML without restarting
  • HTTPS, or a short guide to put Caddy/nginx in front
  • Homebrew
  • Same model name on two hosts without guessing
  • Anthropic tools and images, if anyone uses that path
  • A log of who called which model