Architectural Design, Not Only Model Intelligence, Governs Multi-Agent LLM Performance
cs.AI
Submitted: 2026-02-03
Updated: 2026-09-17
Comments: This paper is accepted to SIGMOD 2027
Code: https://github.com/CoDS-GCS/MAFBench
Project page: https://openai.github.io/openai-agents-python
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Phi-4 Technical Report
- Training Verifiers to Solve Math Word Problems
- DeepSeek-V3 Technical Report
- The Llama 3 Herd of Models
- AgentsNet: Coordination and Collaborative Reasoning in Multi-Agent LLMs
- Adaptive Graph Pruning for Multi-Agent Communication
- GPT-4 Technical Report
- Beyond Pipelines: A Survey of the Paradigm Shift toward Model-Native Agentic AI
- Retrieval Models Aren't Tool-Savvy: Benchmarking Tool Retrieval for Large Language Models
- Generative agent-based modeling with actions grounded in physical, social, or digital space using Concordia
Related papers
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
- MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
- Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems
- VSAL: A Vision Solver with Adaptive Layouts for Graph Property Detection