异构多智能体系统统一提示-模型联合优化框架 UniMASO
原标题：Not Every Agent Needs the Strongest Model: Unified Prompt-Model Optimization for Heterogeneous LLM-based Multi-Agent Systems
AI 摘要
该论文提出 UniMASO，一个面向异构 LLM 多智能体系统的统一优化框架，联合调整各智能体的提示词与执行模型分配，而非像以往工作那样固定其中一项。方法通过局部、下游与全局反馈评估候选更新，结合成本感知的帕累托搜索与拓扑感知协同演化，并用跨层轨迹记忆指导后续搜索。在五个基准和五种执行拓扑上，UniMASO 在集中式执行下将平均分从 69.79 提升至 70.40，同时相比使用 MASPO 的 DeepSeek-V4-Pro 配置降低 85.2% 的 token 加权部署成本，代码已开源。
正文节选
Not Every Agent Needs the Strongest Model: Unified Prompt–Model Optimization for Heterogeneous LLM-based Multi-Agent Systems Abstract Large language model (LLM)-based multi-agent systems (MAS) rely on role-specific prompts and execution models to solve complex tasks collaboratively. However, prompt optimization typically keeps model assignments fixed, leaving differences in role requirements and their associated inference costs outside the search. We present UniMASO, a Unified framework for hete