Models

List Models

Retrieve a list of all models available through Router9.

GET /v1/models

Example Request

curl https://api.router9.com/v1/models \
  -H "Authorization: Bearer sk-r9k-your-key"

Response

{
  "object": "list",
  "data": [
    {
      "id": "anthropic/claude-opus-4.8",
      "name": "anthropic/claude-opus-4.8",
      "architecture": {
        "input_modalities": ["text", "image", "file"],
        "output_modalities": ["text"]
      },
      "context_length": 1000000,
      "created": 1781409302
    },
    {
      "id": "google/gemini-3.5-flash",
      "name": "google/gemini-3.5-flash",
      "architecture": {
        "input_modalities": ["text", "image", "audio"],
        "output_modalities": ["text"]
      },
      "context_length": 1048576,
      "created": 1781409302
    }
  ]
}

Each entry exposes id, name, an architecture block (input_modalities / output_modalities), context_length, and a created timestamp. The exact set of models depends on your server's configuration — always read the live list from this endpoint rather than hardcoding IDs.

Model Routing

Router9 uses a dynamic routing system. When you request a model:

1. The model ID is matched against configured routes using prefix matching
2. Routes are sorted by priority (highest first)
3. The request is forwarded to the matched upstream provider

The auto Model

If your administrator has configured an auto route, setting model to "auto" lets Router9 pick a model for you (weighted across the configured upstreams). auto is not a built-in alias — when no auto route is configured, requesting it returns 404 model_not_found. Check GET /v1/models to see whether auto is available on your deployment.

Available Providers

Providers and models are configured per deployment. A typical hosted catalog looks like:

The exact list of available models depends on your server's configuration. Use the List Models endpoint to see what's currently available.

Model Access by Plan

Calling a model not flagged as free-tier from a FREE account returns 403 model_requires_upgrade. The Free-tier allowlist is managed per model route in the admin dashboard.

Current Limitations

Router9 routes chat / tool-calling completion models only. Two things to know when wiring up a harness:

• Model ids are provider-prefixed slugs, not short aliases. Use the exact ids from GET /v1/models (e.g. anthropic/claude-opus-4.8, google/gemini-3.5-flash). Names like gpt-4o or claude-sonnet-4-… are not aliases — they return 404 model_not_found unless an administrator has configured a matching route. Likewise, auto only works if an auto route exists on your deployment.
• No embeddings endpoint. There is no /v1/embeddings route. Harnesses that need embeddings (for memory/RAG, e.g. Letta) must point their embedding handle at a different provider — only chat models route through Router9.