Skip to main content

Router builder

Use Chat for AI on Dracon. Use this optional page only when you want to draft and export an ordered provider/model list for LiteLLM, OpenCode, or your own client elsewhere. This page does not call providers or run failover.

Loading

Fetching your saved routes…

What it builds

A fallback list is an optional ordered list of providers drawn from your connected keys. The first hop is the primary and later hops are fallbacks for a tool you run elsewhere. Pick providers from the keys you've already added, move them into order, and export the list in the format your chosen tool understands. Onsite AI happens in Chat.

The Dracon optional fallback-list builder: an ordered provider list with an export format picker

Export formats

If you choose to use the list elsewhere, export as opencode.json, LiteLLM YAML, an OpenAI-compatible shell snippet, a raw JSON manifest, or a .env block. This is optional; you do not need an export to use onsite Chat.

Show all export formats
  • LiteLLM YAML — a true cascading chain: one local endpoint, automatic failover across every hop.
  • opencode.json — every hop wired as a provider so you can switch between them in the UI.
  • OpenAI-compatible snippet — a shell setup for any client that speaks the OpenAI API shape.
  • Raw JSON manifest — the route list itself, for anything else you want to wire.
  • .env — the key material for the route, generated locally; configs only ever reference env vars.

Your accounts, your prices

Dracon Chat is the onsite experience: you connect your own provider account and the browser calls that provider directly. This optional list builder runs in your browser and does not proxy or meter model traffic. You keep your provider account and pay its listed price — many have real free tiers. Manage keys in the BYOK provider keys; browse all 26 providers; or read the OpenRouter alternatives comparison.

How it works

  1. Connect provider keys in the BYOK provider keys.
  2. Use Chat directly on Dracon — no export required.
  3. Only if needed, create an optional fallback list and stack providers in priority order.
  4. Copy or download the list for OpenCode, LiteLLM, or your own client elsewhere.