StreamMA 论文提出多智能体推理流式通信，降低延迟并提升准确率

View PDF

Abstract:Multi-agent reasoning systems adopt a generate-then-transfer paradigm that forces end-to-end latency to scale linearly with pipeline depth. We introduce StreamMA, a multi-agent reasoning system that streams each reasoning step to downstream agents as soon as it is generated, pipelining adjacent agents and thus reducing latency. Surprisingly, this pipelining also improves effectiveness: because multi-step reasoning quality is non-uniform and early steps are more reliable than later ones, working with these reliable early steps instead of the full chain prevents error-prone late steps from misleading downstream agents. We formalize both advantages with the first closed-form joint analysis of stream, serial, and single protocols, deriving the effectiveness ordering, speedup upper bound, and cost ratio. Across eight reasoning benchmarks spanning mathematics, science, and code, three frontier LLMs (Claude Opus 4.6, GPT-5.4, and GLM-5.2), and three topologies (Chain, Tree, Graph), StreamMA outperforms both baselines (avg.+7.3pp, max +22.4pp on HMMT2026; Claude Opus 4.6-High). Beyond these contributions, we discover a step-level scaling law: increasing per-agent steps consistently improves both effectiveness and efficiency, a new scaling dimension orthogonal to and composable with agent-count scaling.

Comments:	project page: this https URL
Subjects:	Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
Cite as:	arXiv:2606.05158 [cs.CL]
	(or arXiv:2606.05158v3 [cs.CL] for this version)
	https://doi.org/10.48550/arXiv.2606.05158 arXiv-issued DOI via DataCite

Submission history

From: Zhen Yang [view email]
[v1] Wed, 3 Jun 2026 17:57:04 UTC (139 KB)
[v2] Sat, 1 Aug 2026 06:36:12 UTC (142 KB)
[v3] Tue, 27 Oct 2026 06:32:12 UTC (145 KB)