OpenRouter

OpenRouter is a hosted gateway that fronts hundreds of models (Anthropic, OpenAI, Google, Meta, Mistral, and more) behind a single OpenAI-compatible endpoint and one API key. Declare it once as a model server and any delegator can target the whole catalog.

Prerequisites

An OpenRouter account and an API key from openrouter.ai/keys
An OpenAI-protocol LLM tool — codex works directly; claude/gemini need a bridge (see Protocol compatibility)

Configuration

Export your key (kept out of config — Operator references it by name):

export OPENROUTER_API_KEY="sk-or-..."

Declare the server in operator.toml:

[[model_servers]]
name = "openrouter"
kind = "openrouter"
base_url = "https://openrouter.ai/api/v1"
api_key_env = "OPENROUTER_API_KEY"
display_name = "OpenRouter"

Then reference it from a delegator. The model is an OpenRouter model id (the vendor/model form from its catalog):

[[delegators]]
name = "codex-openrouter-sonnet"
llm_tool = "codex"
model = "anthropic/claude-3.5-sonnet"
model_server = "openrouter"

Listing models

OpenRouter publishes its catalog at /models. Operator probes it for the live model list (which doubles as a reachability check):

# REST
GET /api/v1/model-servers/openrouter/models   # { reachable, models[], error? }

In the VS Code status tree, expand the openrouter server to browse its models; each model’s human label comes from OpenRouter’s name field.

How env injection works

OpenRouter speaks the OpenAI protocol, so when a delegator resolves to it Operator exports:

var	value
`OPENAI_BASE_URL`	`https://openrouter.ai/api/v1`
`OPENAI_API_KEY`	`${OPENROUTER_API_KEY}` (by reference)

The key is injected by reference, never by value — the secret is never written into the on-disk command script. See the Model Providers overview for the full mechanism.