MARS: toward more efficient multi-agent collaboration for LLM reasoning

Abstract

Large language models (LLMs) have achieved impressive results in natural language understanding, yet their reasoning capabilities remain limited when operating as single agents. Multi-Agent Debate (MAD) has been proposed to address this limitation by enabling collaborative reasoning among multiple models in a round-table debate manner. While effective, MAD introduces substantial computational overhead due to the number of agents involved and the frequent communication required. In this paper, we propose MARS (Multi-Agent Review System), a role-based collaboration framework inspired by the review process. In MARS, an author agent generates an initial solution, reviewer agents provide decisions and comments independently, and a meta-reviewer integrates the feedback to make the final decision and guide further revision. This design enhances reasoning quality while avoiding costly reviewer-to-reviewer interactions, thereby controlling token consumption and inference time. We compared MARS with both MAD and other state-of-the-art reasoning strategies across multiple benchmarks. Extensive experiments with different LLMs show that MARS matches the accuracy of MAD while reducing both token usage and inference time by approximately 50\%. Code is available at https://github.com/xwang97/MARS.

Figures (4)

• Figure 1: Overview of the architecture of MARS and baselines. (a) Basic single model inference (top) and self-reflection (bottom). (b) Self-consistency. (c) Multi-Agent-Debate (MAD). (d) Multi-Agent Review System (MARS). In MARS, the author agent receives a user query and generates an initial response. Each reviewer agent evaluates the response and provides a decision, confidence level, and justification (e.g., reasons for the decision, identified author mistakes). The meta-reviewer integrates review comments and makes the final decision, with suggestions for answer revision. Finally, the author agent incorporates the feedback and updates its response, leading to enhanced reasoning.
• Figure 2: Accuracy-Resource trade-off across different models on GPQA. MARS demonstrates a significant reduction in token averaged tokens compared to MAD, while achieving higher accuracy than self-consistency.
• Figure 3: Comparison of MARS and MAD on GPQA with varying number of agents. Row 1: using GPT-3.5-turbo as the backbone; Row 2: using GPT-4o-mini as the backbone. Column 1: accuracy scores; Column 2: averaged number of tokens.
• Figure 4: Case study of MARS on a GSM example. Upon receiving the user query, the author agent first generates an initial response but incorrectly recomputed a given variable, leading to an incorrect final answer. The reviewers identified the mistake and generated feedback, which guided the author agent to revise the solution, resulting in the correct answer.