Routing Overview - Electron Hub API

Every chat, Messages, and Responses request is routed in three layers. Each layer is optional and they combine freely.

Layer	Decides	Set it with	Page
Model selection	Which model answers	`model: "auto"`	Auto Router
Provider selection	Which provider serves the chosen model	`provider` object, `:nitro` / `:floor` / `:exacto` suffixes, `<provider-id>/<model>`	Provider Routing, Model Variants
Fallbacks	Which model runs if the first one cannot be served	`models` (Chat, Responses) or `fallbacks` (Messages)	Model Fallbacks

Without any of these, a request goes to the model you name, served by the cheapest healthy provider that fits the request.

Quick examples

{
  "model": "auto",
  "messages": [{ "role": "user", "content": "Refactor this function to be faster." }]
}

{
  "model": "kimi-k2.6:nitro",
  "messages": [{ "role": "user", "content": "Summarize this log." }]
}

{
  "model": "kimi-k2.6",
  "provider": { "ignore": ["<provider-id>"], "sort": "latency" },
  "messages": [{ "role": "user", "content": "Hello" }]
}

{
  "model": "claude-sonnet-4-5-20250929",
  "models": ["gpt-5.1", "deepseek-v4-flash"],
  "messages": [{ "role": "user", "content": "Hello" }]
}

Which model served the request

The response model field always names the model that produced the answer. Billing uses that model’s price. Auto Router responses also carry the X-EH-Routed-Model header.

Supported endpoints

Endpoint	`auto`	`provider`	Suffixes	Fallbacks
`POST /v1/chat/completions` (and `/v1legacy`, `/v1thinking`)	Yes	Yes	Yes	`models`
`POST /v1/messages`	Yes	Yes	Yes	`fallbacks` or `models`
`POST /v1/responses`	Yes	Yes	Yes	`models`
Embeddings, images, audio, video, moderations	No (`400`)	No	No	No