How it works

• Chat Trigger: Listens for incoming chat messages via the chat trigger node.
• Get Model Names: Calls the “Get Model Names” tool to fetch the latest AI model list from OpenRouter.
• Scrape Capabilities: Uses the “Scrape Model Capabilities” HTTP tool to retrieve a leaderboard of models and their strengths.
• Filter Models: Runs a Code node to filter models by provider, tool support, and structured output capability.
• Select Model: The Model Finding Agent evaluates the user’s request, checks pricing tiers per guidelines, and picks the best-suited model.
• Fetch Model ID: Matches the selected model name to its ID from the fetched list.
• Respond: The Response Agent sends back the chosen model’s name, ID, and a one-line reason.

Set up steps

• Required Credentials: OpenRouter API key for AI Model nodes and Tavily API key for scraping model capabilities.
• Workflow Import: Import the JSON into n8n and enable the “Smart Model Selection” workflow.
• Node Configuration: Attach the Telegram chat trigger (or other chat source), configure Chat credentials, and set up the HTTP Request tool with your Tavily bearer token.
• Approximate setup time: 20–30 minutes to import, configure nodes, and verify tool connections.
• Detailed information with step is added as sticky notes to guide setup.