# Debate Only When Necessary: Adaptive Multiagent Collaboration for Efficient LLM Reasoning

arXiv: https://arxiv.org/abs/2504.05047 · HTML: https://mlsanity.com/paper/2504.05047/

## Key facts

- Cited by 27 papers in the map (data through 14 Oct 2026).
- Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning.
- Builds on 8 of its 32 references that are in the map.
- First posted on arXiv 7 Apr 2025.

## Placed in

- LLM reasoning / [Reasoning for agents, planning and code](https://mlsanity.com/lr/reasoning-for-agents-planning-and-code/)

## Builds on (most cited first, 8 of 8)

- [AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation](https://mlsanity.com/paper/2308.08155/) (2023)
- [Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters](https://mlsanity.com/paper/2408.03314/) (2024)
- [Multi-Agent Collaboration Mechanisms: A Survey of LLMs](https://mlsanity.com/paper/2501.06322/) (2025)
- [AI Agents That Matter](https://mlsanity.com/paper/2407.01502/) (2024)
- [Confidence Improves Self-Consistency in LLMs](https://mlsanity.com/paper/2502.06233/) (2025)
- [Many Heads Are Better Than One: Improved Scientific Idea Generation by A LLM-Based Multi-Agent System](https://mlsanity.com/paper/2410.09403/) (2024)
- [Rethinking Mixture-of-Agents: Is Mixing Different Large Language Models Beneficial?](https://mlsanity.com/paper/2502.00674/) (2025)
- [CER: Confidence Enhanced Reasoning in LLMs](https://mlsanity.com/paper/2502.14634/) (2025)

## Built on by (most cited first, 27 of 27)

- [Can LLM Agents Really Debate? A Controlled Study of Multi-Agent Debate in Logical Reasoning](https://mlsanity.com/paper/2511.07784/) (2025)
- [Graph-Augmented Large Language Model Agents: Current Progress and Future Prospects](https://mlsanity.com/paper/2507.21407/) (2025)
- [Aware First, Think Less: Dynamic Boundary Self-Awareness Drives Extreme Reasoning Efficiency in Large Language Models](https://mlsanity.com/paper/2508.11582/) (2025)
- [AgentArk: Distilling Multi-Agent Intelligence into a Single LLM Agent](https://mlsanity.com/paper/2602.03955/) (2026)
- [How to Train a Leader: Hierarchical Reasoning in Multi-Agent LLMs](https://mlsanity.com/paper/2507.08960/) (2025)
- [Reasoning Models Generate Societies of Thought](https://mlsanity.com/paper/2601.10825/) (2026)
- [Eigen-1: Adaptive Multi-Agent Refinement with Monitor-Based RAG for Scientific Reasoning](https://mlsanity.com/paper/2509.21193/) (2025)
- [iMAD: Intelligent Multi-Agent Debate for Efficient and Accurate LLM Inference](https://mlsanity.com/paper/2511.11306/) (2025)
- [LLM-First Search: Self-Guided Exploration of the Solution Space](https://mlsanity.com/paper/2506.05213/) (2025)
- [Epistemic Gain, Aleatoric Cost: Uncertainty Decomposition in Multi-Agent Debate for Math Reasoning](https://mlsanity.com/paper/2603.01221/) (2026)
- [Emergence of Biased Consensus in Multi-Agent LLM Debates](https://mlsanity.com/paper/2608.02827/) (2026)
- [The Effectiveness of Approximate Regularized Replay for Efficient Supervised Fine-Tuning of Large Language Models](https://mlsanity.com/paper/2512.22337/) (2025)
- [EMS: Multi-Agent Voting via Efficient Majority-then-Stopping](https://mlsanity.com/paper/2604.02863/) (2026)
- [Social Dynamics as Critical Vulnerabilities that Undermine Objective Decision-Making in LLM Collectives](https://mlsanity.com/paper/2604.06091/) (2026)
- [HCP-MAD:Heterogeneous Consensus-Progressive Reasoning for Efficient Multi-Agent Debate](https://mlsanity.com/paper/2604.09679/) (2026)
- [Share More, Search Less: Collaborative Parallel Thinking for Efficient Test-Time Scaling](https://mlsanity.com/paper/2605.27030/) (2026)
- [LLM Agents for Deliberative Collaboration: A Study on Joint Decision Making Under Partial Observability](https://mlsanity.com/paper/2607.06157/) (2026)
- [CRAwDAD: Causal Reasoning Augmentation with Dual-Agent Debate](https://mlsanity.com/paper/2511.22854/) (2025)
- [SVR-MAD: A Bayesian-Inspired Framework for Posterior-Guided Multi-Agent Debate](https://mlsanity.com/paper/2605.23099/) (2026)
- [FaLCon: Facet-Anchored Retrieval with Late Consensus for Sim2Real Text-Based Person Anomaly Search](https://mlsanity.com/paper/2608.09474/) (2026)
- [Statistical Scouting Finds Debate-Safe but Not Debate-Useful Cases: A Matched-Ceiling Study of Open-Weight LLM Reasoning Protocols](https://mlsanity.com/paper/2605.09618/) (2026)
- [Confidence Composition for Multiagent Language Model Systems](https://mlsanity.com/paper/2606.13591/) (2026)
- [Group Perspective Matters: Regulating Debate Relationships Can Mitigate Blind Conformity in Multi-Agent Debate](https://mlsanity.com/paper/2608.03648/) (2026)
- [Semantic Uncertainty-Guided Orchestration in Hierarchical Multi-Agent Systems](https://mlsanity.com/paper/2608.14707/) (2026)
- [From Process Loss to Assembly Bonus: Human-Grounded Diagnosis of Multi-Agent LLM Collaboration](https://mlsanity.com/paper/2609.13261/) (2026)
- [Measuring Collapse and Correction in Homogeneous-Panel LLM Debate](https://mlsanity.com/paper/2609.35279/) (2026)
- [BaRe-Mem: Bayesian Reliability Memory for Robust and Adaptive Agent Consultation](https://mlsanity.com/paper/2609.35551/) (2026)

## Abstract (arXiv)

Multiagent collaboration has emerged as a promising framework for enhancing the reasoning capabilities of large language models (LLMs). Despite improvements in reasoning, the approach introduces substantial computational overhead resulting from iterative agent interactions. Furthermore, engaging in unnecessary debates increases the risk of generating erroneous responses. To address these challenges, we propose Debate Only When Necessary (DOWN), an adaptive multiagent debate framework that selectively activates debate based on the confidence score of the agent's initial response. Debate is activated only for queries requiring further deliberation, during which agents refine their outputs by referencing peer responses and associated confidence scores. Evaluations on benchmarks show that DOWN improves efficiency by up to six times while preserving or even outperforming the performance of existing methods. Further analysis indicates that DOWN effectively mitigates the risk of error propagation stemming from the unnecessary debate process. These findings demonstrate the effectiveness of our approach in delivering high-performance LLM solutions at a lower computational cost.