Category:

Evaluation & Reliability

Definition

Using multiple models to vote or agree on an answer.

Explanation

Cross‑model consensus reduces hallucinations by generating answers from multiple LLMs and selecting the most consistent output. This technique works like an ensemble: different models bring diverse reasoning paths, and consensus selects the most stable, reliable result.

Technical Architecture

Input → Multiple LLMs → Consensus Engine → Verified Output

Core Component

Voting algorithm, agreement scorer, reranker

Use Cases

Legal QA, safety‑critical tasks, analytics, summarization

Pitfalls

High compute cost; models may agree on wrong answers

LLM Keywords

LLM Consensus, Multi-model Agreement, Ensemble Llm

Related Concepts

Related Frameworks

• Verification Layers
• Self‑Consistency
• Evaluation

• Ensemble Reasoning Architecture

Back to Glossary Index