Token-Efficient Multi-Agent Collaboration via System One-Guided Computational Division of Labor

Zihan Zhou; Xinzhe Hu; Hanxu Yang; Liangjian Wen; Zhao Kang

cs.MAOct 6, 2026

Token-Efficient Multi-Agent Collaboration via System One-Guided Computational Division of Labor

Authors: Zihan Zhou, Xinzhe Hu, Hanxu Yang, Liangjian Wen, Zhao Kang

Organizations: University of Electronic Science and Technology of China · Southwestern University of Finance and Economics

Abstract

Large language model (LLM)-based multi-agent systems (MAS) have become a promising paradigm for complex information-seeking and reasoning tasks by enabling collaborative problem solving among specialized agents. However, existing MAS frameworks tightly couple task reasoning with coordination operations, including task selection, role assignment, message routing, and context management. As interactions grow, using powerful LLMs for these bounded control decisions introduces substantial token overhead and latency, limiting the scalability of agentic Web services. In this paper, we investigate whether coordination can be decoupled from expensive reasoning without compromising collaborative performance. We propose S1-MAS, a token-efficient multi-agent framework based on System One-guided computational division of labor. S1-MAS assigns bounded coordination decisions to lightweight System One models while reserving open-ended reasoning for capable LLM workers. Specifically, a lightweight controller selects inspection conditions, chooses subsequent tasks, and determines termination, while a compact reader retrieves condition-relevant evidence from authorized sources to support these decisions. Through a decision-evidence loop, selected tasks dynamically determine worker roles and source access, enabling adaptive collaboration without task-specific training. Extensive experiments on seven diverse benchmarks demonstrate that S1-MAS achieves superior accuracy while substantially reducing the inference cost. Across individual comparisons with AgentVerse, DyLAN, and SelfOrg on seven benchmarks, S1-MAS reduces GPT-4o token consumption by 44.9%-97.2% and measured end-to-end latency by 37.8%-93.0%. These results highlight its potential for scalable and cost-effective agentic Web applications.

Figures & tables

Figure 1: S1-MAS decouples coordination from reasoning: a System One controller makes bounded coordination decisions, while a lightweight reader localizes relevant evidence, reducing token overhead while preserving LLM reasoning capability.

Figure 2: Overview of S1-MAS. (a) The state

S_{t}

records the history of

t

completed worker executions; in this AQuA example, the latest response

m_{t}

generated by task

j_{t}

serves as the current candidate answer. (b) A lightweight System One controller makes bounded coordination decisions by selecting inspection conditions, choosing the next task, or terminating based on evidence localized by a lightweight reader. (c) When collaboration continues, an LLM worker executes

j_{t+1}

to generate

m_{t+1}

and update the state to

S_{t+1}

; otherwise, the controller returns an archived answer upon termination.

Direct Answer
Method	MMLU	GSM8K	AQuA	MultiArith	SVAMP	HumanEval	Average
Vanilla	85.19 $\pm$ 1.00	94.93 $\pm$ 1.33	75.86 $\pm$ 4.34	98.81 $\pm$ 1.36	93.72 $\pm$ 0.26	85.67 $\pm$ 3.44	89.03 $\pm$ 1.26
Single-Agent Prompting
CoT	85.40 $\pm$ 1.64	97.13 $\pm$ 0.32	76.17 $\pm$ 1.87	99.23 $\pm$ 0.81	93.75 $\pm$ 0.81	87.60 $\pm$ 4.60	89.88 $\pm$ 0.97
SC-CoT	89.11 $\pm$ 1.00	97.30 $\pm$ 0.10	82.40 $\pm$ 3.32	99.40 $\pm$ 1.03	94.41 $\pm$ 1.73	88.98 $\pm$ 1.26	91.93 $\pm$ 0.57
Fixed Multi-Agent Topologies

Table 1: Performance comparison using GPT-4o. Accuracy values are reported as percentages. Results are means with standard deviations over three seeds. Average summarizes performance across six datasets. Best and second-best means are bold and underlined , respectively.

Method	MMLU		GSM8K		AQuA		MultiArith		SVAMP		HumanEval		Average
	Ptok.	Ctok.	Ptok.	Ctok.	Ptok.	Ctok.	Ptok.	Ctok.	Ptok.	Ctok.	Ptok.	Ctok.	Ptok.	Ctok.
Vanilla	117	25	80	98	105	176	62	48	62	39	161	155	98	90
MAD-M 2 (S)	4,066	1,407	2,907	1,235	4,764	2,045	2,164	729	2,052	697	6,418	2,734	3,728	1,475
SelfOrg	9,795	5,429	7,032	3,254	10,910	5,979	4,682	1,922	4,835	2,097	17,513	10,909	9,128	4,932
AgentVerse	7,018	3,562	6,780	3,205	9,645	5,206	4,156	1,688	4,332	1,826	14,070	7,728	7,667	3,869
MACNET	2,356	1,165	2,329	1,270	3,472	2,097	1,511	722	1,539	745	3,984	2,272	2,532	1,379

Table 2: Mean GPT-4o tokens per question, averaged over three seeds under the same configurations as Table 1 (lower is better). Ptok. and Ctok. denote input and generation tokens across all calls. Values are rounded to integers.

Figure 3: Mean end-to-end latency across six benchmarks, averaged over three seeds under the same configurations as Tables 1 and 2 . Lower is better; the star highlights S1-MAS.

Method	MMLU	AQuA	HumanEval	Average	GPT tokens
S1-MAS	80.17	81.15	91.18	84.17	535,918
w/ Random	78.43	78.50	90.08	82.34	542,491
w/ GPT	78.87	79.91	90.63	83.14	2,190,250
w/o Reader	79.08	78.35	89.81	82.41	482,314

Table 3: Ablation results with GPT-4o-mini, averaged over three seeds. Accuracy (%) is reported on MMLU, AQuA, and HumanEval. GPT tokens is the per-seed total over these three test sets, averaged across seeds.

Method	EM (%)	GPT tokens	Latency (s)
AgentVerse	89.78	32,386	28.78
DyLAN	89.89	9,927	15.22
SelfOrg	89.11	28,229	61.03
S1-MAS	90.11	4,847	9.47

Table 4: WebSRC performance and efficiency with GPT-4o. Exact match (EM) scores are averaged across three random seeds. GPT tokens and latency are per-question means over the same runs. Best results are bold.

Explore similar work

Phase-Scheduled Multi-Agent Systems for Token-Efficient Coordination
Apr 19, 2026Mohit DubeyModel-Based Multi-Agent Systems Global Coordination
Open Gigantic
CONCAT: Consensus- and Confidence-Driven Ad Hoc Teaming for Efficient LLM-Based Multi-Agent Systems
May 28, 2026Ziyang Ma, Dingyi Zhang, Sichu Liang +4Mixture-Of-Agents Multi-Agent Communication
Southeast University · Huawei Technologies Ltd
Decentralized Multi-Agent Systems with Shared Context
Jun 9, 2026Yuzhen Mao, Jerry Gu, Aadi Chauhan +3Multi-Agent Communication Test-Time Scaling
Stanford University

Apr 19, 2026, cs.AIJ/K move · Enter open · S save

Phase-Scheduled Multi-Agent Systems for Token-Efficient Coordination

Mohit Dubey

Open Gigantic

Model-Based Multi-Agent Systems Global Coordination

Open paper AskPDF arXiv

Multi-agent systems (MAS) powered by large language models suffer from severe token inefficiency arising from two compounding sources: (i) unstructured parallel execution, where all agents activate simultaneously irrespective of input readiness; and (ii) unrestricted context sharing, where every agent receives the full accumulated context regardless of relevance. Existing mitigation strategies - static pruning, hierarchical decomposition, and learned routing - treat coordination as a structural allocation problem and fundamentally ignore its temporal dimension. We propose Phase-Scheduled Multi-Agent Systems (PSMAS), a framework that reconceptualizes agent activation as continuous control over a shared attention space modeled on a circular manifold. Each agent i is assigned a fixed angular phase theta_i in the range [0, 2*pi], derived from the task dependency topology; a global sweep signal phi(t) rotates at velocity omega, activating only agents within an angular window epsilon. Idle agents receive compressed context summaries, reducing per-step token consumption. We implement PSMAS on LangGraph, evaluate on four structured benchmarks (HotPotQA-MAS, HumanEval-MAS, ALFWorld-Multi, WebArena-Coord) and two unstructured conversational settings, and prove stability, convergence, and optimality results for the sweep dynamics. PSMAS achieves a mean token reduction of 27.3 percent (range 21.4-34.8 percent) while maintaining task performance within 2.1 percentage points of a fully activated baseline (p < 0.01, n = 500 per configuration), and outperforms the strongest learned routing baseline by 5.6 percentage points in token reduction with 2.0 percentage points less performance drop. Crucially, we show that scheduling and compression are independent sources of gain: scheduling alone accounts for 18-20 percentage points of reduction, robust to compression degradation up to alpha = 0.40.

Abstract

Explore similar work

Phase-Scheduled Multi-Agent Systems for Token-Efficient Coordination

CONCAT: Consensus- and Confidence-Driven Ad Hoc Teaming for Efficient LLM-Based Multi-Agent Systems

Decentralized Multi-Agent Systems with Shared Context