Skip to main content

OpenRouterModel

What it is​

A small wrapper that constructs a langchain_openai.ChatOpenAI client configured to call the OpenRouter API endpoint.

Public API​

  • class OpenRouterModel
    • __init__(api_key: str)
      • Stores the OpenRouter API key and sets the fixed base_url to https://openrouter.ai/api/v1.
    • get_model(model_id: str) -> langchain_openai.ChatOpenAI
      • Returns a ChatOpenAI instance configured with:
        • model=model_id
        • api_key=SecretStr(self.api_key)
        • base_url=self.base_url

Configuration/Dependencies​

  • Dependencies:
    • langchain_openai.ChatOpenAI
    • pydantic.SecretStr
  • Configuration:
    • Requires an OpenRouter API key passed to OpenRouterModel(api_key=...).
    • Uses a hardcoded base URL: https://openrouter.ai/api/v1.

Usage​

from naas_abi_marketplace.applications.openrouter.models.OpenRouterModel import OpenRouterModel

router = OpenRouterModel(api_key="YOUR_OPENROUTER_API_KEY")
llm = router.get_model("openai/gpt-4o-mini")  # model_id is passed through as-is

# llm is a langchain_openai.ChatOpenAI instance configured for OpenRouter

Caveats​

  • No validation is performed for api_key or model_id; invalid values will fail when the returned client is used.
  • The base URL is fixed in code and cannot be overridden via the public API.