Reference

API Reference

Every endpoint, parameter and error the API returns.

Base URL

base url

https://reroute.wtf/api/v1

Every response is JSON. Errors share one shape: an error object with a numeric code equal to the HTTP status and a human-readable message.

Chat completions

POST /api/v1/chat/completions — OpenAI Chat Completions compatible. Requires a bearer key.

Request body

Field	Type	Description
`messages`	array, required	Conversation so far. Must be non-empty, otherwise 400.
`model`	string	Catalog model ID. Omitted or `reroute/auto` uses `openai/gpt-oss-20b`.
`stream`	boolean	Server-sent events when `true`. Default `false`.
`any other field`	—	Forwarded to the carrier unchanged: `temperature`, `max_tokens`, `top_p`, `tools`, `response_format`, `stop`, … Check a model's `supported_parameters`.

request

POST https://reroute.wtf/api/v1/chat/completions
Authorization: Bearer <REROUTE_API_KEY>
Content-Type: application/json

{
  "model": "anthropic/claude-sonnet-4.5",
  "messages": [
    { "role": "system", "content": "You are a helpful assistant." },
    { "role": "user", "content": "Explain request routing in one line." }
  ],
  "temperature": 0.7,
  "max_tokens": 256,
  "stream": false
}

Response

The upstream completion, with id replaced by an Reroute generation ID, model set to the model you asked for, provider naming the carrier that served it, and usage.cost set to the USD deducted from your balance.

200 OK

{
  "id": "gen-3y8Rq0vZ1kXbT2mN4pLc",
  "object": "chat.completion",
  "created": 1791411439,
  "model": "anthropic/claude-sonnet-4.5",
  "provider": "Anthropic",
  "choices": [
    {
      "index": 0,
      "message": { "role": "assistant", "content": "Moving goods from where they are to where they're needed." },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 24,
    "completion_tokens": 14,
    "total_tokens": 38,
    "cost": 0.000282
  }
}

Streaming

With stream: true the response is text/event-stream. Each data: line is a chunk with the same id, model and provider. The last chunk before data: [DONE] carries usage when the carrier reports it. The generation ID is also sent as the X-Generation-Id header so you can look the request up before the stream ends.

stream

HTTP/1.1 200 OK
Content-Type: text/event-stream; charset=utf-8
X-Generation-Id: gen-3y8Rq0vZ1kXbT2mN4pLc

data: {"id":"gen-3y8Rq0vZ1kXbT2mN4pLc","model":"anthropic/claude-sonnet-4.5","provider":"Anthropic","choices":[{"index":0,"delta":{"content":"Moving"}}]}

data: {"id":"gen-3y8Rq0vZ1kXbT2mN4pLc","model":"anthropic/claude-sonnet-4.5","provider":"Anthropic","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":24,"completion_tokens":14}}

data: [DONE]

Errors

error shape

{
  "error": {
    "code": 402,
    "message": "Insufficient credits. Add more using https://reroute.wtf/settings/credits"
  }
}

Status	When
400	Body is not JSON; `messages` missing or empty; model ID not in the catalog.
401	No key, unknown key, or the key is disabled.
402	Paid model with a balance at or below zero, or the key has reached its credit limit.
404	Generation, model or route not found.
429	The carrier rate-limited the request. Retry with backoff.
502	The carrier returned an error or did not respond. The message names the carrier's reason when it gives one.
503	No carrier can serve this model right now. Nothing is charged.

Requests that fail are recorded in your activity log with a cost of $0.

Get a generation

GET /api/v1/generation?id=<gen-id> — metadata for one of your own requests. Latency is time to first token in milliseconds; generation_time is the full request duration.

shell

curl "https://reroute.wtf/api/v1/generation?id=gen-3y8Rq0vZ1kXbT2mN4pLc" \
  -H "Authorization: Bearer $REROUTE_API_KEY"

{
  "data": {
    "id": "gen-3y8Rq0vZ1kXbT2mN4pLc",
    "model": "anthropic/claude-sonnet-4.5",
    "provider_name": "Anthropic",
    "app": "Your App",
    "streamed": true,
    "created_at": "2026-10-08T00:17:19.148Z",
    "latency": 612,
    "generation_time": 1840,
    "tokens_prompt": 24,
    "tokens_completion": 14,
    "native_tokens_reasoning": 0,
    "total_cost": 0.000282,
    "finish_reason": "stop",
    "status": 200
  }
}

Get current key

GET /api/v1/key — label, usage, limit, remaining limit, free-tier flag and disabled flag for the key you authenticate with. See Authentication.

Get credits

GET /api/v1/credits — returns { "data": { "total_credits": number, "total_usage": number } } in USD for the key's account.

List models

GET /api/v1/models — public. The full catalog. GET /api/v1/models/<author>/<slug>/endpoints — public. The carriers serving one model, with pricing, context, quantization and uptime.

List providers

GET /api/v1/providers — public. Every provider with headquarters, datacenters and policy links.