n8n Ships Multi-Model Debate Template for AI Consensus Answers.

n8n has released a workflow template that turns a single question into a structured debate between multiple AI models. The template, featured this week on the platform's official template library, orchestrates four models to answer independently, critique each other's responses, and ultimately produce a single consensus-driven answer.

The workflow is inspired by Andrej Karpathy's LLM Council but rebuilt natively in n8n's visual automation environment.

The template follows a three-stage process. First, four different models — Gemini, Llama, Gemma, and Mistral by default — each generate an independent response to the user's question. No model sees the others' answers at this stage.

Next comes peer review. Each model receives all four responses and evaluates them, identifying strengths, weaknesses, and potential blind spots. This cross-examination step surfaces gaps or overconfidence in individual answers.

Finally, a fifth model, DeepSeek R1, reads all peer reviews and synthesizes a refined answer that incorporates the strongest elements from each response.

n8n positions the template for users who need to pressure-test ideas from multiple angles. The company suggests use cases including prepping for meetings with stakeholders who hold different views, identifying blind spots in reasoning, and improving reliability of AI outputs by forcing models to critique one another.

The workflow is configurable. Users can swap in different models or adjust parameters to suit their needs. The template uses OpenRouter as the LLM provider, requiring API credits and an n8n credential setup.

The approach addresses a known limitation of single-model outputs: overconfidence and hallucination. By forcing models to review each other's reasoning, the workflow surfaces contradictions and gaps that might otherwise go unnoticed.

This aligns with n8n's recent focus on production-grade AI orchestration. The company published a Production AI Playbook on managing complex multi-agent systems earlier this year and has been steadily adding capabilities for developers building agent-driven workflows.

The template is available now in n8n's workflow library. Setup requires an OpenRouter account and takes under five minutes once credentials are configured. Users interact with the workflow through n8n's Chat Trigger interface.